Re: [E-devel] [PATCH] evas/wayland-egl: Don't crash when re-win is NULL
Eduardo, The patch is simple enough. If it fixes the problem for you, then go ahead and push it. The wayland_egl engine is going to be rewritten at some point (when I find some time) to make use of evas_gl stuff, but if this fixes it for you then feel free to push it. dh On 15/05/13 20:56, Eduardo Lima (Etrunko) wrote: From: Eduardo Lima (Etrunko) eduardo.l...@intel.com Hi Devilhorns, I wanted to send this email to the mailing list before commiting because I am a bit confused of why this is crashing. Take a look on this backtrace http://fpaste.org/12383/64474713/ The scenario is basically the normal of a Ecore_Evas application: ecore_evas_new() - ecore_evas_resize() - ecore_evas_show() - ecore_mainloop_begin(). At some point there is a call to ecore_evas_hide(), and then, when I call ecore_evas_show() again, the application crashes. Regards, Etrunko Signed-off-by: Eduardo Lima (Etrunko) eduardo.l...@intel.com --- src/modules/evas/engines/wayland_egl/evas_engine.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/modules/evas/engines/wayland_egl/evas_engine.c b/src/modules/evas/engines/wayland_egl/evas_engine.c index 0d9761a..0cd3ac6 100644 --- a/src/modules/evas/engines/wayland_egl/evas_engine.c +++ b/src/modules/evas/engines/wayland_egl/evas_engine.c @@ -544,6 +544,7 @@ static int _re_wincheck(Render_Engine *re) { if (!re) return 0; + if (!re-win) return 0; if (re-win-surf) return 1; eng_window_resurf(re-win); if (!re-win-surf) @@ -761,7 +762,7 @@ eng_setup(Evas *eo_e, void *in) if ((re-win) (inc)) re-win-gl_context-references--; } - } + } else if ((re-win-w != e-output.w) || (re-win-h != e-output.h)) { -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Proxy render doesn't work on GL
I just tested your sample but it works fine here. what efl version did u use? I'm using the latest git source. -Regards, Hermet- -Original Message- From: Rafael Antognolliantogno...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-11 (토) 05:47:58 Subject: [E-devel] Proxy render doesn't work on GL Guys, there's a corner case for the proxy render that won't work on GL engines. The test is attached and is very simple. Basically, it creates an Edje object, with a black rectangle and a green rectangle, and this green rectangle has a map applied to it. This Edje object is moved almost entirely off the screen, and a proxy is created to draw this Edje object, but only the green rectangle is seen (not the black one). If the order of the rectangles inside the .edc is changed (the mapped one is created first), things work correctly. If the Edje object starts all visible, it also works. The important part is: I really think that when the surface for the mapped object is created, it is making something wrong with the proxy surface, and resetting its content. After some investigation, I noticed the following scenario, when evas_render is called: 1) proxy object creates a surface for rendering the proxy content; 2) the black rectangle is rendered on this surface; 3) the green rectangle (mapped) creates a new surface for rendering itself; -- I think this is wrong somehow 4) the green rectangle surface is drawn over the proxy surface; 5) the proxy surface is rendered on the screen. Of course there are many other things involved, but I found that these are the most important ones affecting things. If we change the order of things, when the green rectangle surface is created, it can reset the proxy surface content but the black rectangle will be drawn only after that, so everything looks good. If the Edje starts inside the canvas geometry, then the green rectangle surface is created and used before the proxy surface is created, so everything looks ok too. In other words, I think that the following behavior works ok: 1) surface is created 2) surface content is rendered 3) surface is drawn somewhere 4) new surface is created 5) new surface content is rendered 6) new surface is drawn somewhere But the following one does not work: 1) surface is created 2) surface content is rendered 3) new surface is created 4) new surface content is rendered 5) new surface is rendered on the previous one 6) surface is drawn somewhere On the last scheme, even if I remove 4) and 5), just the fact of creating the new surface makes the content of the first one vanish. Please take a look at _pool_tex_render_new() and related functions, the bug may be somewhere there. Thanks, -- Rafael Antognolli -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Evas Image Bug
On Thu, 16 May 2013 14:52:52 +0900 (KST) ChunEon Park her...@naver.com said: i was wondering if it was the same bug... the cedric bug... :) it looks it's fixed by raster. :) -Regards, Hermet- -Original Message- From: Bluezeryohpo...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-15 (수) 00:44:48 Subject: [E-devel] Evas Image Bug Hello, Recently I found a evas bug. Sometimes it can not show the image correctly. I used bisect to find the bug. bisect said that below commit will be the bug. commit c3f9d3b8b8f2530923353beff4e67661a75f7614 Author: Cedric Bail cedric.bail@samsung.com Date: Tue May 7 10:50:57 2013 +0900 evas: Make Evas_Loader API public. Cedric may be the person :) I attached sample application to generate this bug. In genlist, image on first item can not be shown. Please help me. -- BRs, Kim. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property
On Thu, 16 May 2013 14:45:16 +0900 (KST) ChunEon Park her...@naver.com said: interesting... it was working fine last saturday for me on my s9... :/ series 9. -Regards, Hermet- -Original Message- From: Carsten Haitzlerras...@rasterman.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: ChunEon Parkher...@naver.com; Sent: 2013-05-16 (목) 14:28:00 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On Thu, 16 May 2013 13:46:00 +0900 (KST) ChunEon Park hermet@naver.com said: samsung p580? By the way, my laptop doesn't worked on backlight control. Do you have any idea to check the problem? -Regards, Hermet- -Original Message- From: Boris Faureboris@fau.re To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-13 (월) 20:21:21 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On 13-05-13 11:02, Michael Blumenkrantz wrote: hmmm sounds like the issue should be fixed elsewhere instead. can you tell me what values are being passed to that function? I'll have a look in _bl_sys_find() in src/bin/e_backlight.c: eeze_udev_find_by_filter(backlight, NULL, NULL); -- Boris Faure Pointer Arithmetician -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property
But, it didn't work at all when i installed e17 on my s9 first time. -Regards, Hermet- -Original Message- From: Carsten Haitzlerras...@rasterman.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: ChunEon Parkher...@naver.com; Sent: 2013-05-16 (목) 15:40:41 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On Thu, 16 May 2013 14:45:16 +0900 (KST) ChunEon Park hermet@naver.com said: interesting... it was working fine last saturday for me on my s9... :/ series 9. -Regards, Hermet- -Original Message- From: Carsten Haitzlerraster@rasterman.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: ChunEon Parkhermet@naver.com; Sent: 2013-05-16 (목) 14:28:00 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On Thu, 16 May 2013 13:46:00 +0900 (KST) ChunEon Park hermet@naver.com said: samsung p580? By the way, my laptop doesn't worked on backlight control. Do you have any idea to check the problem? -Regards, Hermet- -Original Message- From: Boris Faureboris@fau.re To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-13 (월) 20:21:21 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On 13-05-13 11:02, Michael Blumenkrantz wrote: hmmm sounds like the issue should be fixed elsewhere instead. can you tell me what values are being passed to that function? I'll have a look in _bl_sys_find() in src/bin/e_backlight.c: eeze_udev_find_by_filter(backlight, NULL, NULL); -- Boris Faure Pointer Arithmetician -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property
On Thu, 16 May 2013 15:56:30 +0900 (KST) ChunEon Park her...@naver.com said: really weird... it has worked solidly for me for e17 and later (e18 git). But, it didn't work at all when i installed e17 on my s9 first time. -Regards, Hermet- -Original Message- From: Carsten Haitzlerras...@rasterman.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: ChunEon Parkher...@naver.com; Sent: 2013-05-16 (목) 15:40:41 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On Thu, 16 May 2013 14:45:16 +0900 (KST) ChunEon Park hermet@naver.com said: interesting... it was working fine last saturday for me on my s9... :/ series 9. -Regards, Hermet- -Original Message- From: Carsten Haitzlerraster@rasterman.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: ChunEon Parkhermet@naver.com; Sent: 2013-05-16 (목) 14:28:00 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On Thu, 16 May 2013 13:46:00 +0900 (KST) ChunEon Park hermet@naver.com said: samsung p580? By the way, my laptop doesn't worked on backlight control. Do you have any idea to check the problem? -Regards, Hermet- -Original Message- From: Boris Faureboris@fau.re To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-13 (월) 20:21:21 Subject: Re: [E-devel] [EGIT] [core/efl] master 01/01: eeze: do no try to match an unknown property On 13-05-13 11:02, Michael Blumenkrantz wrote: hmmm sounds like the issue should be fixed elsewhere instead. can you tell me what values are being passed to that function? I'll have a look in _bl_sys_find() in src/bin/e_backlight.c: eeze_udev_find_by_filter(backlight, NULL, NULL); -- Boris Faure Pointer Arithmetician -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com
Re: [E-devel] Problem with evas 1.7 on ARM
On Tue, May 07, 2013 at 05:23:40PM +0100, Daniel Willmann wrote: On 07/05/13 17:12, Martin Jansa wrote: On Tue, May 07, 2013 at 10:54:24AM +0100, Daniel Willmann wrote: On 05/05/13 09:11, Jérôme Pinot wrote: Anyway, it becomes obvious that evas was the actual culprit as reverting it for the 1.7.5 version makes e17 happy again. So I started to bisect the damn thing which proves to be quite painful under qemu, even with the help of distcc. And just because it doesn't happend so much these days: THANK YOU GIT FOR YOUR MARVELLOUS BISECT FEATURE! So the commit that gives problem is: 8 From 9a7a28b7582f516c67b23c4bb1a016cfc2ec5b31 Mon Sep 17 00:00:00 2001 From: Stefan Schmidt ste...@datenfreihafen.org Date: Fri, 08 Feb 2013 16:02:51 + Subject: Backport rev 83789 SVN revision: 83793 8 Link: http://git.enlightenment.org/legacy/evas.git/commit/?id=9a7a28b7582f516c67 b23c4bb1a016cfc2ec5b31 Ouch, that should have been reverted some time ago as it caused other issues as well (and wasn't actually a fix). I reverted it now. Did you forget to push it? I did push it. gitk shows me it's upstream I don't see it at http://git.enlightenment.org/legacy/evas.git/log/ Maybe another effect of the diskspace-issue earlier? I just cloned the repo and it's definitely there. I will check about cgit. In evas-1.7 branch? I still don't see it there.. $ git log origin/evas-1.7 -- src/modules/engines/software_x11/evas_xlib_outbuf.c | head commit 9a7a28b7582f516c67b23c4bb1a016cfc2ec5b31 Author: Stefan Schmidt ste...@datenfreihafen.org Date: Fri Feb 8 16:02:51 2013 + Backport rev 83789 SVN revision: 83793 Thanks Some people on IRC told me they were already using last git on ARM without trouble, so I wonder. Maybe the gcc optimizations? I build on ARM using -march=armv4 -mtune=xscale. Hmm, it would only cause problems if you are using the evas rotation support so maybe that's the difference? Daniel -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Martin 'JaMa' Jansa jabber: martin.ja...@gmail.com signature.asc Description: Digital signature -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Problem with evas 1.7 on ARM
On 16/05/13 09:36, Martin Jansa wrote: On Tue, May 07, 2013 at 05:23:40PM +0100, Daniel Willmann wrote: On 07/05/13 17:12, Martin Jansa wrote: I don't see it at http://git.enlightenment.org/legacy/evas.git/log/ Maybe another effect of the diskspace-issue earlier? I just cloned the repo and it's definitely there. I will check about cgit. In evas-1.7 branch? I still don't see it there.. Haha! Layer 8 issue, I still had my migration repo set as remote. $ git log origin/evas-1.7 -- src/modules/engines/software_x11/evas_xlib_outbuf.c | head commit 9a7a28b7582f516c67b23c4bb1a016cfc2ec5b31 Author: Stefan Schmidt ste...@datenfreihafen.org Date: Fri Feb 8 16:02:51 2013 + Backport rev 83789 SVN revision: 83793 It is fixed now. http://git.enlightenment.org/legacy/evas.git/ shows it. Thanks for checking again, I would have forgotten! Regards, Daniel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Problem with evas 1.7 on ARM
On Thu, May 16, 2013 at 10:27:22AM +0100, Daniel Willmann wrote: On 16/05/13 09:36, Martin Jansa wrote: On Tue, May 07, 2013 at 05:23:40PM +0100, Daniel Willmann wrote: On 07/05/13 17:12, Martin Jansa wrote: I don't see it at http://git.enlightenment.org/legacy/evas.git/log/ Maybe another effect of the diskspace-issue earlier? I just cloned the repo and it's definitely there. I will check about cgit. In evas-1.7 branch? I still don't see it there.. Haha! Layer 8 issue, I still had my migration repo set as remote. $ git log origin/evas-1.7 -- src/modules/engines/software_x11/evas_xlib_outbuf.c | head commit 9a7a28b7582f516c67b23c4bb1a016cfc2ec5b31 Author: Stefan Schmidt ste...@datenfreihafen.org Date: Fri Feb 8 16:02:51 2013 + Backport rev 83789 SVN revision: 83793 It is fixed now. http://git.enlightenment.org/legacy/evas.git/ shows it. Thanks for checking again, I would have forgotten! pity it didn't make it to 1.7.7 release :/ Rafael: would you mind spinning 1.7.7.1 with this? -- Martin 'JaMa' Jansa jabber: martin.ja...@gmail.com signature.asc Description: Digital signature -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] elua bitching on exit
2013/5/3 David Seikel onef...@gmail.com On Thu, 2 May 2013 18:24:43 +0200 Leif Middelschulte leif.middelschu...@gmail.com wrote: Hi, I've written an 'all script' lua edje, which creates/modifies/deletes a couple of objects. Everything seems fine, until the application is closed (not crashing), when it complains about exceeding lua's memory limit. Is this expected? If not, do you need me to provide code, or is it even a known issue? Don't think I've seen that before. Please provide code. Edje Lua tracks how much memory you have allocated to each Lua script, and will complain about that if you exceed it, but that should not be happening when you are shutting down. Sorry for taking this long. See the attached edc. If you confirm this, I can open a ticket, if necessary. Thanks for looking into this. -- A big old stinking pile of genius that no one wants coz there are too many silver coated monkeys in the world. -- Get 100% visibility into Java/.NET code with AppDynamics Lite It's a free troubleshooting tool designed for production Get down to code-level detail for bottlenecks, with 2% overhead. Download for free and get started troubleshooting in minutes. http://p.sf.net/sfu/appdyn_d2d_ap2 ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Leif lua_leak.edc Description: Binary data -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Proxy render doesn't work on GL
I'm using latest git too. Have you tried it with ECORE_EVAS_ENGINE=opengl_x11 set? On Thu, May 16, 2013 at 3:37 AM, ChunEon Park her...@naver.com wrote: I just tested your sample but it works fine here. what efl version did u use? I'm using the latest git source. -Regards, Hermet- -Original Message- From: Rafael Antognolliantogno...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-11 (토) 05:47:58 Subject: [E-devel] Proxy render doesn't work on GL Guys, there's a corner case for the proxy render that won't work on GL engines. The test is attached and is very simple. Basically, it creates an Edje object, with a black rectangle and a green rectangle, and this green rectangle has a map applied to it. This Edje object is moved almost entirely off the screen, and a proxy is created to draw this Edje object, but only the green rectangle is seen (not the black one). If the order of the rectangles inside the .edc is changed (the mapped one is created first), things work correctly. If the Edje object starts all visible, it also works. The important part is: I really think that when the surface for the mapped object is created, it is making something wrong with the proxy surface, and resetting its content. After some investigation, I noticed the following scenario, when evas_render is called: 1) proxy object creates a surface for rendering the proxy content; 2) the black rectangle is rendered on this surface; 3) the green rectangle (mapped) creates a new surface for rendering itself; -- I think this is wrong somehow 4) the green rectangle surface is drawn over the proxy surface; 5) the proxy surface is rendered on the screen. Of course there are many other things involved, but I found that these are the most important ones affecting things. If we change the order of things, when the green rectangle surface is created, it can reset the proxy surface content but the black rectangle will be drawn only after that, so everything looks good. If the Edje starts inside the canvas geometry, then the green rectangle surface is created and used before the proxy surface is created, so everything looks ok too. In other words, I think that the following behavior works ok: 1) surface is created 2) surface content is rendered 3) surface is drawn somewhere 4) new surface is created 5) new surface content is rendered 6) new surface is drawn somewhere But the following one does not work: 1) surface is created 2) surface content is rendered 3) new surface is created 4) new surface content is rendered 5) new surface is rendered on the previous one 6) surface is drawn somewhere On the last scheme, even if I remove 4) and 5), just the fact of creating the new surface makes the content of the first one vanish. Please take a look at _pool_tex_render_new() and related functions, the bug may be somewhere there. Thanks, -- Rafael Antognolli -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Rafael Antognolli -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
Please revert this !!! and re-apply WITHOUT the formatting changes dh On 16/05/13 12:18, J챕r챕my Zurcher - Enlightenment Git wrote: jeyzu pushed a commit to branch master. commit ab505d5d23299e658b39fa7d0f1aa73742e3e7fb Author: Jérémy Zurcher jer...@asynk.ch Date: Wed May 15 16:42:31 2013 +0200 ecore evas wayland: if(()) are used for assignments, add missing NULL fields --- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c | 4 ++-- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c | 8 +--- 2 files changed, 7 insertions(+), 5 deletions(-) diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c index 3aa60bf..820f001 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c @@ -435,7 +435,7 @@ _ecore_evas_wl_alpha_set(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -463,7 +463,7 @@ _ecore_evas_wl_transparent_set(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c index 2bac6f1..93e3361 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c @@ -76,7 +76,9 @@ static Ecore_Evas_Engine_Func _ecore_wl_engine_func = NULL, // focus skip set NULL, //_ecore_evas_wl_common_render, _ecore_evas_wl_common_screen_geometry_get, - _ecore_evas_wl_common_screen_dpi_get + _ecore_evas_wl_common_screen_dpi_get, + NULL, // func msg parent send + NULL // func msg send }; /* external variables */ @@ -458,7 +460,7 @@ _ecore_evas_wayland_shm_alpha_do(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -507,7 +509,7 @@ _ecore_evas_wayland_shm_transparent_do(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Proxy render doesn't work on GL
And what you mean by it works? Is the black rectangle also appearing on the test, or only a small green rectangle on the white background? The problem here is that the black rectangle does not appear because of the green rectangle. On Thu, May 16, 2013 at 8:30 AM, Rafael Antognolli antogno...@gmail.com wrote: I'm using latest git too. Have you tried it with ECORE_EVAS_ENGINE=opengl_x11 set? On Thu, May 16, 2013 at 3:37 AM, ChunEon Park her...@naver.com wrote: I just tested your sample but it works fine here. what efl version did u use? I'm using the latest git source. -Regards, Hermet- -Original Message- From: Rafael Antognolliantogno...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-11 (토) 05:47:58 Subject: [E-devel] Proxy render doesn't work on GL Guys, there's a corner case for the proxy render that won't work on GL engines. The test is attached and is very simple. Basically, it creates an Edje object, with a black rectangle and a green rectangle, and this green rectangle has a map applied to it. This Edje object is moved almost entirely off the screen, and a proxy is created to draw this Edje object, but only the green rectangle is seen (not the black one). If the order of the rectangles inside the .edc is changed (the mapped one is created first), things work correctly. If the Edje object starts all visible, it also works. The important part is: I really think that when the surface for the mapped object is created, it is making something wrong with the proxy surface, and resetting its content. After some investigation, I noticed the following scenario, when evas_render is called: 1) proxy object creates a surface for rendering the proxy content; 2) the black rectangle is rendered on this surface; 3) the green rectangle (mapped) creates a new surface for rendering itself; -- I think this is wrong somehow 4) the green rectangle surface is drawn over the proxy surface; 5) the proxy surface is rendered on the screen. Of course there are many other things involved, but I found that these are the most important ones affecting things. If we change the order of things, when the green rectangle surface is created, it can reset the proxy surface content but the black rectangle will be drawn only after that, so everything looks good. If the Edje starts inside the canvas geometry, then the green rectangle surface is created and used before the proxy surface is created, so everything looks ok too. In other words, I think that the following behavior works ok: 1) surface is created 2) surface content is rendered 3) surface is drawn somewhere 4) new surface is created 5) new surface content is rendered 6) new surface is drawn somewhere But the following one does not work: 1) surface is created 2) surface content is rendered 3) new surface is created 4) new surface content is rendered 5) new surface is rendered on the previous one 6) surface is drawn somewhere On the last scheme, even if I remove 4) and 5), just the fact of creating the new surface makes the content of the first one vanish. Please take a look at _pool_tex_render_new() and related functions, the bug may be somewhere there. Thanks, -- Rafael Antognolli -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Rafael Antognolli -- Rafael Antognolli -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial.
Re: [E-devel] [EGIT] [core/efl] master 02/05: eo ptr ind: mostly cosmetic
Hey, Thanks for your efforts, but if possible please avoid such changes in the future. 1. Don't mix cosmetic changes with actual code affecting changes. 2. The cosmetic changes you've made are disputable, and anyhow just make it very annoying to revert old patches, use git blame to find who changed a line, and etc. -- Tom. On 16/05/13 12:18, Jérémy Zurcher - Enlightenment Git wrote: jeyzu pushed a commit to branch master. commit a9e69d519cd5240320e278cdc4efa8b1723a5fa1 Author: Jérémy Zurcher jer...@asynk.ch Date: Thu May 16 10:55:32 2013 +0200 eo ptr ind: mostly cosmetic - add and use SHIFT_* macros - rename queue into fifo - try to clarify the structure top table - mid table - table[entry] --- src/lib/eo/eo_ptr_indirection.c | 216 +--- 1 file changed, 111 insertions(+), 105 deletions(-) diff --git a/src/lib/eo/eo_ptr_indirection.c b/src/lib/eo/eo_ptr_indirection.c index 74a3926..47c5295 100644 --- a/src/lib/eo/eo_ptr_indirection.c +++ b/src/lib/eo/eo_ptr_indirection.c @@ -17,14 +17,16 @@ * the Eo object to supply a better memory management by preventing bad usage * of the pointers. * - * Eo * is no more a pointer but an index to an entry into a ids table. - * For a better memory usage, we don't allocate all the tables at the beginning, - * but only when needed (i.e no more empty entries in allocated tables. - * In addition, tables are composed of intermediate tables, this for memory - * optimizations. Finding the different table, intermediate table and relative - * entry is done by bits manipulation of the id: + * Eo * is no more a pointer but indexes to an entry into an ids table. + * For a better memory usage: + * - a tree structure is used, composed of a top level table pointing at + * mid tables pointing at tables composed of entries. + * - tables are allocated when needed (i.e no more empty entries in allocated tables. + * For now there is no mechanism to free empty tables. * - * id = Table | Inter_table | Entry | Generation + * An Eo id is contructed by bits manipulation of table indexes and a generation. + * + * id = Mid Table | Table | Entry | Generation * * Generation helps finding abuse of ids. When an entry is assigned to an * object, a generation is inserted into the id. If the developer uses this id @@ -32,69 +34,73 @@ * entry of the table, the generation will be different and an error will * occur when accessing with the old id. * - * Each table is composed of: - * - an index 'start' indicating which entry is the next one to use. - * - 2 indexes 'queue_head' and 'queue_tail' defining a queue (fifo), + * Each Table is composed of: + * - an index 'start' indicating which free entry is the next one to use. + * - 2 indexes 'fifo_head' and 'fifo_tail' defining a fifo, * that will help us to store the entries to be reused. It stores only the * entries that have been used at least one time. The entries that have * never been used are pointed by the start parameter. * - entries composed of: *- a pointer to the object + *- an index 'next_in_fifo' used to chain the free entries in the fifo *- a flag indicating if the entry is active *- a generation assigned to the object - *- an index 'next_in_queue' used to chain the entries in the queue * * When an entry is searched into a table, we first use one of the entries that - * has never been used. If there is none, we try to pop from the queue. + * has never been used. If there is none, we try to pop from the fifo. + * If a such entry doesn't exist, we pass to the next table. + * When an entry is found, we reserve it to the object pointer + * then contruct and return the related Eo id. + * * Assigning all the entries of a table before trying to reuse them from * the fifo ensures that we are not going to soon recycle a released entry, * thus minimize the risks of an aggressive del() then use() on a single entry. - * If a such entry doesn't exist, we pass to the next table. - * When an entry is found, we reserve it to the object pointer and create - * the id with the table id, the intermediate table id, the entry and a - * generation. + * * The indexes and a reference to the last table which served an entry is kept * and is reused prior to the others untill it is full. * When an object is freed, the entry into the table is released by appending - * it to the queue. + * it to the fifo. */ #if SIZEOF_UINTPTR_T == 4 /* 32 bits */ -# define BITS_FOR_IDS_TABLE 5 -# define BITS_FOR_IDS_INTER_TABLE 5 -# define BITS_FOR_ID_IN_TABLE12 -# define BITS_FOR_GENERATION_COUNTER 10 +# define BITS_MID_TABLE_ID5 +# define BITS_TABLE_ID5 +# define BITS_ENTRY_ID 12 +# define BITS_GENERATION_COUNTER 10 typedef int16_t Table_Index;
Re: [E-devel] [PATCH] evas/wayland-egl: Don't crash when re-win is NULL
On Thu, May 16, 2013 at 3:01 AM, Christopher Michael cp.mich...@samsung.com wrote: Eduardo, The patch is simple enough. If it fixes the problem for you, then go ahead and push it. The wayland_egl engine is going to be rewritten at some point (when I find some time) to make use of evas_gl stuff, but if this fixes it for you then feel free to push it. Alright, I'm pushing it later today. I was wondering if this was an actual problem with the engine or something I was doing wrong, but looking at the code, there are actually some valid cases where re-win can be NULL, so that should be checked. Regards, Etrunko dh On 15/05/13 20:56, Eduardo Lima (Etrunko) wrote: From: Eduardo Lima (Etrunko) eduardo.l...@intel.com Hi Devilhorns, I wanted to send this email to the mailing list before commiting because I am a bit confused of why this is crashing. Take a look on this backtrace http://fpaste.org/12383/64474713/ The scenario is basically the normal of a Ecore_Evas application: ecore_evas_new() - ecore_evas_resize() - ecore_evas_show() - ecore_mainloop_begin(). At some point there is a call to ecore_evas_hide(), and then, when I call ecore_evas_show() again, the application crashes. Regards, Etrunko Signed-off-by: Eduardo Lima (Etrunko) eduardo.l...@intel.com --- src/modules/evas/engines/wayland_egl/evas_engine.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/modules/evas/engines/wayland_egl/evas_engine.c b/src/modules/evas/engines/wayland_egl/evas_engine.c index 0d9761a..0cd3ac6 100644 --- a/src/modules/evas/engines/wayland_egl/evas_engine.c +++ b/src/modules/evas/engines/wayland_egl/evas_engine.c @@ -544,6 +544,7 @@ static int _re_wincheck(Render_Engine *re) { if (!re) return 0; + if (!re-win) return 0; if (re-win-surf) return 1; eng_window_resurf(re-win); if (!re-win-surf) @@ -761,7 +762,7 @@ eng_setup(Evas *eo_e, void *in) if ((re-win) (inc)) re-win-gl_context-references--; } - } + } else if ((re-win-w != e-output.w) || (re-win-h != e-output.h)) { -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Eduardo de Barros Lima ◤✠◢ ebl...@gmail.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Proxy render doesn't work on GL
ahhh ok. i see the problem when i launch it on the opengl_x11 . now the black rect is invisible before first timer cb is called. -Regards, Hermet- -Original Message- From: Rafael Antognolliantogno...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-16 (목) 21:00:33 Subject: Re: [E-devel] Proxy render doesn't work on GL And what you mean by it works? Is the black rectangle also appearing on the test, or only a small green rectangle on the white background? The problem here is that the black rectangle does not appear because of the green rectangle. On Thu, May 16, 2013 at 8:30 AM, Rafael Antognolli antognolli@gmail.com wrote: I'm using latest git too. Have you tried it with ECORE_EVAS_ENGINE=opengl_x11 set? On Thu, May 16, 2013 at 3:37 AM, ChunEon Park hermet@naver.com wrote: I just tested your sample but it works fine here. what efl version did u use? I'm using the latest git source. -Regards, Hermet- -Original Message- From: Rafael Antognolliantognolli@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-11 (토) 05:47:58 Subject: [E-devel] Proxy render doesn't work on GL Guys, there's a corner case for the proxy render that won't work on GL engines. The test is attached and is very simple. Basically, it creates an Edje object, with a black rectangle and a green rectangle, and this green rectangle has a map applied to it. This Edje object is moved almost entirely off the screen, and a proxy is created to draw this Edje object, but only the green rectangle is seen (not the black one). If the order of the rectangles inside the .edc is changed (the mapped one is created first), things work correctly. If the Edje object starts all visible, it also works. The important part is: I really think that when the surface for the mapped object is created, it is making something wrong with the proxy surface, and resetting its content. After some investigation, I noticed the following scenario, when evas_render is called: 1) proxy object creates a surface for rendering the proxy content; 2) the black rectangle is rendered on this surface; 3) the green rectangle (mapped) creates a new surface for rendering itself; -- I think this is wrong somehow 4) the green rectangle surface is drawn over the proxy surface; 5) the proxy surface is rendered on the screen. Of course there are many other things involved, but I found that these are the most important ones affecting things. If we change the order of things, when the green rectangle surface is created, it can reset the proxy surface content but the black rectangle will be drawn only after that, so everything looks good. If the Edje starts inside the canvas geometry, then the green rectangle surface is created and used before the proxy surface is created, so everything looks ok too. In other words, I think that the following behavior works ok: 1) surface is created 2) surface content is rendered 3) surface is drawn somewhere 4) new surface is created 5) new surface content is rendered 6) new surface is drawn somewhere But the following one does not work: 1) surface is created 2) surface content is rendered 3) new surface is created 4) new surface content is rendered 5) new surface is rendered on the previous one 6) surface is drawn somewhere On the last scheme, even if I remove 4) and 5), just the fact of creating the new surface makes the content of the first one vanish. Please take a look at _pool_tex_render_new() and related functions, the bug may be somewhere there. Thanks, -- Rafael Antognolli -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Rafael
Re: [E-devel] [PATCH] evas/wayland-egl: Don't crash when re-win is NULL
On 16/05/13 13:43, Eduardo Lima (Etrunko) wrote: On Thu, May 16, 2013 at 3:01 AM, Christopher Michael cp.mich...@samsung.com wrote: Eduardo, The patch is simple enough. If it fixes the problem for you, then go ahead and push it. The wayland_egl engine is going to be rewritten at some point (when I find some time) to make use of evas_gl stuff, but if this fixes it for you then feel free to push it. Alright, I'm pushing it later today. I was wondering if this was an actual problem with the engine Aside from being a bloody mess you mean ?? ;) or something I was doing wrong, but looking at the code, there are actually some valid cases where re-win can be NULL, so that should be checked. Indeed. No worries with a push ;) dh Regards, Etrunko dh On 15/05/13 20:56, Eduardo Lima (Etrunko) wrote: From: Eduardo Lima (Etrunko) eduardo.l...@intel.com Hi Devilhorns, I wanted to send this email to the mailing list before commiting because I am a bit confused of why this is crashing. Take a look on this backtrace http://fpaste.org/12383/64474713/ The scenario is basically the normal of a Ecore_Evas application: ecore_evas_new() - ecore_evas_resize() - ecore_evas_show() - ecore_mainloop_begin(). At some point there is a call to ecore_evas_hide(), and then, when I call ecore_evas_show() again, the application crashes. Regards, Etrunko Signed-off-by: Eduardo Lima (Etrunko) eduardo.l...@intel.com --- src/modules/evas/engines/wayland_egl/evas_engine.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/modules/evas/engines/wayland_egl/evas_engine.c b/src/modules/evas/engines/wayland_egl/evas_engine.c index 0d9761a..0cd3ac6 100644 --- a/src/modules/evas/engines/wayland_egl/evas_engine.c +++ b/src/modules/evas/engines/wayland_egl/evas_engine.c @@ -544,6 +544,7 @@ static int _re_wincheck(Render_Engine *re) { if (!re) return 0; + if (!re-win) return 0; if (re-win-surf) return 1; eng_window_resurf(re-win); if (!re-win-surf) @@ -761,7 +762,7 @@ eng_setup(Evas *eo_e, void *in) if ((re-win) (inc)) re-win-gl_context-references--; } - } + } else if ((re-win-w != e-output.w) || (re-win-h != e-output.h)) { -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Proxy render doesn't work on GL
Do you need help for debugging this? I can help with more info if needed, but not sure how much it would help... On Thu, May 16, 2013 at 9:43 AM, ChunEon Park her...@naver.com wrote: ahhh ok. i see the problem when i launch it on the opengl_x11 . now the black rect is invisible before first timer cb is called. -Regards, Hermet- -Original Message- From: Rafael Antognolliantogno...@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-16 (목) 21:00:33 Subject: Re: [E-devel] Proxy render doesn't work on GL And what you mean by it works? Is the black rectangle also appearing on the test, or only a small green rectangle on the white background? The problem here is that the black rectangle does not appear because of the green rectangle. On Thu, May 16, 2013 at 8:30 AM, Rafael Antognolli antognolli@gmail.com wrote: I'm using latest git too. Have you tried it with ECORE_EVAS_ENGINE=opengl_x11 set? On Thu, May 16, 2013 at 3:37 AM, ChunEon Park hermet@naver.com wrote: I just tested your sample but it works fine here. what efl version did u use? I'm using the latest git source. -Regards, Hermet- -Original Message- From: Rafael Antognolliantognolli@gmail.com To: Enlightenment developer listenlightenment-devel@lists.sourceforge.net; Cc: Sent: 2013-05-11 (토) 05:47:58 Subject: [E-devel] Proxy render doesn't work on GL Guys, there's a corner case for the proxy render that won't work on GL engines. The test is attached and is very simple. Basically, it creates an Edje object, with a black rectangle and a green rectangle, and this green rectangle has a map applied to it. This Edje object is moved almost entirely off the screen, and a proxy is created to draw this Edje object, but only the green rectangle is seen (not the black one). If the order of the rectangles inside the .edc is changed (the mapped one is created first), things work correctly. If the Edje object starts all visible, it also works. The important part is: I really think that when the surface for the mapped object is created, it is making something wrong with the proxy surface, and resetting its content. After some investigation, I noticed the following scenario, when evas_render is called: 1) proxy object creates a surface for rendering the proxy content; 2) the black rectangle is rendered on this surface; 3) the green rectangle (mapped) creates a new surface for rendering itself; -- I think this is wrong somehow 4) the green rectangle surface is drawn over the proxy surface; 5) the proxy surface is rendered on the screen. Of course there are many other things involved, but I found that these are the most important ones affecting things. If we change the order of things, when the green rectangle surface is created, it can reset the proxy surface content but the black rectangle will be drawn only after that, so everything looks good. If the Edje starts inside the canvas geometry, then the green rectangle surface is created and used before the proxy surface is created, so everything looks ok too. In other words, I think that the following behavior works ok: 1) surface is created 2) surface content is rendered 3) surface is drawn somewhere 4) new surface is created 5) new surface content is rendered 6) new surface is drawn somewhere But the following one does not work: 1) surface is created 2) surface content is rendered 3) new surface is created 4) new surface content is rendered 5) new surface is rendered on the previous one 6) surface is drawn somewhere On the last scheme, even if I remove 4) and 5), just the fact of creating the new surface makes the content of the first one vanish. Please take a look at _pool_tex_render_new() and related functions, the bug may be somewhere there. Thanks, -- Rafael Antognolli -- Learn Graph Databases - Download FREE O'Reilly Book Graph Databases is the definitive new guide to graph databases and their applications. This 200-page book is written by three acclaimed leaders in the field. The early access version is available now. Download your free book today! http://p.sf.net/sfu/neotech_d2d_may ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
sorry, but what are those formatting changes ?? … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. On Thursday 16 May 2013 12:59, Christopher Michael wrote : Please revert this !!! and re-apply WITHOUT the formatting changes dh On 16/05/13 12:18, J챕r챕my Zurcher - Enlightenment Git wrote: jeyzu pushed a commit to branch master. commit ab505d5d23299e658b39fa7d0f1aa73742e3e7fb Author: Jérémy Zurcher jer...@asynk.ch Date: Wed May 15 16:42:31 2013 +0200 ecore evas wayland: if(()) are used for assignments, add missing NULL fields --- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c | 4 ++-- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c | 8 +--- 2 files changed, 7 insertions(+), 5 deletions(-) diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c index 3aa60bf..820f001 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c @@ -435,7 +435,7 @@ _ecore_evas_wl_alpha_set(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -463,7 +463,7 @@ _ecore_evas_wl_transparent_set(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c index 2bac6f1..93e3361 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c @@ -76,7 +76,9 @@ static Ecore_Evas_Engine_Func _ecore_wl_engine_func = NULL, // focus skip set NULL, //_ecore_evas_wl_common_render, _ecore_evas_wl_common_screen_geometry_get, - _ecore_evas_wl_common_screen_dpi_get + _ecore_evas_wl_common_screen_dpi_get, + NULL, // func msg parent send + NULL // func msg send }; /* external variables */ @@ -458,7 +460,7 @@ _ecore_evas_wayland_shm_alpha_do(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -507,7 +509,7 @@ _ecore_evas_wayland_shm_transparent_do(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Jérémy Zurcher av General Guisan 49 1400 Yverdon-les-bains +41 (0) 79 599 84 27 -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) dh On Thursday 16 May 2013 12:59, Christopher Michael wrote : Please revert this !!! and re-apply WITHOUT the formatting changes dh On 16/05/13 12:18, J챕r챕my Zurcher - Enlightenment Git wrote: jeyzu pushed a commit to branch master. commit ab505d5d23299e658b39fa7d0f1aa73742e3e7fb Author: Jérémy Zurcher jer...@asynk.ch Date: Wed May 15 16:42:31 2013 +0200 ecore evas wayland: if(()) are used for assignments, add missing NULL fields --- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c | 4 ++-- src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c | 8 +--- 2 files changed, 7 insertions(+), 5 deletions(-) diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c index 3aa60bf..820f001 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_egl.c @@ -435,7 +435,7 @@ _ecore_evas_wl_alpha_set(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -463,7 +463,7 @@ _ecore_evas_wl_transparent_set(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; diff --git a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c index 2bac6f1..93e3361 100644 --- a/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c +++ b/src/modules/ecore_evas/engines/wayland/ecore_evas_wayland_shm.c @@ -76,7 +76,9 @@ static Ecore_Evas_Engine_Func _ecore_wl_engine_func = NULL, // focus skip set NULL, //_ecore_evas_wl_common_render, _ecore_evas_wl_common_screen_geometry_get, - _ecore_evas_wl_common_screen_dpi_get + _ecore_evas_wl_common_screen_dpi_get, + NULL, // func msg parent send + NULL // func msg send }; /* external variables */ @@ -458,7 +460,7 @@ _ecore_evas_wayland_shm_alpha_do(Ecore_Evas *ee, int alpha) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-alpha == alpha)) return; + if (ee-alpha == alpha) return; ee-alpha = alpha; wdata = ee-engine.data; @@ -507,7 +509,7 @@ _ecore_evas_wayland_shm_transparent_do(Ecore_Evas *ee, int transparent) LOGFN(__FILE__, __LINE__, __FUNCTION__); if (!ee) return; - if ((ee-transparent == transparent)) return; + if (ee-transparent == transparent) return; ee-transparent = transparent; wdata = ee-engine.data; -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 02/05: eo ptr ind: mostly cosmetic
Hi, I will improve, about the 'cosmetic' - there was an acual mix in the comments on table allocations - using entry_id for the index of the entry within the table, but using table_id for the inter table index within the master and inter_table_id for the table index within the inter table … it's messy as entry_id was not to be changed I did what I did but as said, I will improve regards On Thursday 16 May 2013 13:25, Tom Hacohen wrote : Hey, Thanks for your efforts, but if possible please avoid such changes in the future. 1. Don't mix cosmetic changes with actual code affecting changes. 2. The cosmetic changes you've made are disputable, and anyhow just make it very annoying to revert old patches, use git blame to find who changed a line, and etc. -- Tom. On 16/05/13 12:18, Jérémy Zurcher - Enlightenment Git wrote: jeyzu pushed a commit to branch master. commit a9e69d519cd5240320e278cdc4efa8b1723a5fa1 Author: Jérémy Zurcher jer...@asynk.ch Date: Thu May 16 10:55:32 2013 +0200 eo ptr ind: mostly cosmetic - add and use SHIFT_* macros - rename queue into fifo - try to clarify the structure top table - mid table - table[entry] --- src/lib/eo/eo_ptr_indirection.c | 216 +--- 1 file changed, 111 insertions(+), 105 deletions(-) diff --git a/src/lib/eo/eo_ptr_indirection.c b/src/lib/eo/eo_ptr_indirection.c index 74a3926..47c5295 100644 --- a/src/lib/eo/eo_ptr_indirection.c +++ b/src/lib/eo/eo_ptr_indirection.c @@ -17,14 +17,16 @@ * the Eo object to supply a better memory management by preventing bad usage * of the pointers. * - * Eo * is no more a pointer but an index to an entry into a ids table. - * For a better memory usage, we don't allocate all the tables at the beginning, - * but only when needed (i.e no more empty entries in allocated tables. - * In addition, tables are composed of intermediate tables, this for memory - * optimizations. Finding the different table, intermediate table and relative - * entry is done by bits manipulation of the id: + * Eo * is no more a pointer but indexes to an entry into an ids table. + * For a better memory usage: + * - a tree structure is used, composed of a top level table pointing at + * mid tables pointing at tables composed of entries. + * - tables are allocated when needed (i.e no more empty entries in allocated tables. + * For now there is no mechanism to free empty tables. * - * id = Table | Inter_table | Entry | Generation + * An Eo id is contructed by bits manipulation of table indexes and a generation. + * + * id = Mid Table | Table | Entry | Generation * * Generation helps finding abuse of ids. When an entry is assigned to an * object, a generation is inserted into the id. If the developer uses this id @@ -32,69 +34,73 @@ * entry of the table, the generation will be different and an error will * occur when accessing with the old id. * - * Each table is composed of: - * - an index 'start' indicating which entry is the next one to use. - * - 2 indexes 'queue_head' and 'queue_tail' defining a queue (fifo), + * Each Table is composed of: + * - an index 'start' indicating which free entry is the next one to use. + * - 2 indexes 'fifo_head' and 'fifo_tail' defining a fifo, * that will help us to store the entries to be reused. It stores only the * entries that have been used at least one time. The entries that have * never been used are pointed by the start parameter. * - entries composed of: *- a pointer to the object + *- an index 'next_in_fifo' used to chain the free entries in the fifo *- a flag indicating if the entry is active *- a generation assigned to the object - *- an index 'next_in_queue' used to chain the entries in the queue * * When an entry is searched into a table, we first use one of the entries that - * has never been used. If there is none, we try to pop from the queue. + * has never been used. If there is none, we try to pop from the fifo. + * If a such entry doesn't exist, we pass to the next table. + * When an entry is found, we reserve it to the object pointer + * then contruct and return the related Eo id. + * * Assigning all the entries of a table before trying to reuse them from * the fifo ensures that we are not going to soon recycle a released entry, * thus minimize the risks of an aggressive del() then use() on a single entry. - * If a such entry doesn't exist, we pass to the next table. - * When an entry is found, we reserve it to the object pointer and create - * the id with the table id, the intermediate table id, the entry and a - * generation. + * * The indexes and a reference to the last table which served an entry is kept * and is reused
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:30, Tom Hacohen wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. In this Particular issue, no. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. Well, opinions are like everybody has one (I won't finish that sentence) -- Tom. dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On Thu, May 16, 2013 at 6:54 PM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) I am actually interested in this sentence because I am going to spend my time to fix things in elementary. I know there are many bugs in elementary but can you describe how/why/what/which is ugly in elementary code? This might be super helpful to me when I fix/refactor elementary. Thanks. Daniel Juyung Seo (SeoZ) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Thanks. Daniel Juyung Seo (SeoZ) -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) dh Thanks. Daniel Juyung Seo (SeoZ) -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 14:41, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 6:54 PM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) I am actually interested in this sentence because I am going to spend my time to fix things in elementary. I know there are many bugs in elementary but can you describe how/why/what/which is ugly in elementary code? This might be super helpful to me when I fix/refactor elementary. Thanks. Daniel Juyung Seo (SeoZ) I am not going to explain the ugliness of the elementary code on the mailing list as that is just going to start Format Wars: Episode 9000 but if you are asking my personal opinion, I will gladly send that in another mail. dh and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
[E-devel] Disk selector is in such a bad case it's not even funny
Maintainers of this stupid widget, please fix it. Ways to reproduce: elementary_test Disk selector See how slow, stupid and full of errors it is. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Disk selector is in such a bad case it's not even funny
On 16/05/13 16:11, Tom Hacohen wrote: Maintainers of this stupid widget, please fix it. Ways to reproduce: elementary_test Disk selector See how slow, stupid and full of errors it is. Bad *state. :) I'm also in a bad case, hm... state. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. -- Rafael Antognolli -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Our coding guidelines don't really say anything regarding over resynthesizing, but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? +1 Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. Sure, I totally agree here...but of course others will disagree saying it's ugly ... can't win either way ;) dh -- Rafael Antognolli -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
[E-devel] A hell lot of Conditional jump or move depends on uninitialised value
Yo, valgrind elementary_test List When starting: ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x58776FA: _evas_image_file_header (evas_image_load.c:191) ==20532==by 0x5877BCF: evas_common_load_rgba_image_module_from_file (evas_image_load.c:292) ==20532==by 0x583D9ED: _evas_cache_image_entry_new (evas_cache_image.c:279) ==20532==by 0x583F091: evas_cache_image_request (evas_cache_image.c:832) ==20532==by 0x587956B: evas_common_load_image_from_file (evas_image_main.c:766) ==20532==by 0x58A2FA3: eng_image_load (evas_engine.c:957) ==20532==by 0x57B9FA7: _image_file_set (evas_object_image.c:662) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) ==20532==by 0x64900D1: _eo_dov_internal (eo.c:400) ==20532==by 0x649030D: eo_do_internal (eo.c:429) ==20532==by 0x57B9D62: evas_object_image_file_set (evas_object_image.c:633) ==20532==by 0x4F54D1B: _elm_image_smart_file_set (elm_image.c:923) After scrolling a bit (all the way and back) and pressing the Next API function a few times (scroll all the way!) I get a storm of errors, for example: ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x57D0550: evas_object_render_pre_effect_updates (evas_object_main.c:470) ==20532==by 0x57C8839: evas_object_image_render_pre (evas_object_image.c:4375) ==20532==by 0x582C422: _evas_render_phase1_direct (evas_render.c:326) ==20532==by 0x5830840: evas_render_updates_internal (evas_render.c:1585) ==20532==by 0x58328BC: _canvas_render_async (evas_render.c:2067) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) ==20532==by 0x64900D1: _eo_dov_internal (eo.c:400) ==20532==by 0x649030D: eo_do_internal (eo.c:429) ==20532==by 0x5832823: evas_render_async (evas_render.c:2057) ==20532==by 0x141F8683: _ecore_evas_x_render (ecore_evas_x.c:661) ==20532==by 0x5D989F3: _ecore_evas_idle_enter (ecore_evas.c:98) ==20532==by 0x5B64964: _ecore_call_task_cb (ecore_private.h:303) ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x582B663: evas_object_clip_recalc (evas_inline.x:294) ==20532==by 0x582C74D: _evas_render_phase1_object_process (evas_render.c:408) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582D047: _evas_render_phase1_process (evas_render.c:669) ==20532==by 0x5830714: evas_render_updates_internal (evas_render.c:1565) ==20532==by 0x58328BC: _canvas_render_async (evas_render.c:2067) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) To me it smells like a non-potty trained CoW. But it might be something else. Completely up to date with everything (as of this moment). -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] A hell lot of Conditional jump or move depends on uninitialised value
On 16/05/13 17:06, Tom Hacohen wrote: Yo, valgrind elementary_test List When starting: [...] After scrolling a bit (all the way and back) and pressing the Next API function a few times (scroll all the way!) I get a storm of errors, for example: [...] I can reproduce this as well. It usually happens when I click and try to scroll while the previous scrolling animation is still running. Regards, Daniel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) dh -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] A hell lot of Conditional jump or move depends on uninitialised value
On 16/05/13 17:06, Tom Hacohen wrote: Yo, valgrind elementary_test List When starting: ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x58776FA: _evas_image_file_header (evas_image_load.c:191) ==20532==by 0x5877BCF: evas_common_load_rgba_image_module_from_file (evas_image_load.c:292) ==20532==by 0x583D9ED: _evas_cache_image_entry_new (evas_cache_image.c:279) ==20532==by 0x583F091: evas_cache_image_request (evas_cache_image.c:832) ==20532==by 0x587956B: evas_common_load_image_from_file (evas_image_main.c:766) ==20532==by 0x58A2FA3: eng_image_load (evas_engine.c:957) ==20532==by 0x57B9FA7: _image_file_set (evas_object_image.c:662) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) ==20532==by 0x64900D1: _eo_dov_internal (eo.c:400) ==20532==by 0x649030D: eo_do_internal (eo.c:429) ==20532==by 0x57B9D62: evas_object_image_file_set (evas_object_image.c:633) ==20532==by 0x4F54D1B: _elm_image_smart_file_set (elm_image.c:923) First one is actually super consistent and mega-bad opening a ticket regarding this one. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 17:17, Chris Michael wrote: On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) Because: 1. Compiling without warnings helps assuring our users that the software we produce is of high quality (yes, clang warnings count). 2. I don't want this ugly epidemic to spread. :) 3. I don't see you do the same in other pieces of code you write. And last, but not least: the guidelines about following the surrounding code are about adding code, not re-factoring. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 17:29, Tom Hacohen wrote: On 16/05/13 17:17, Chris Michael wrote: On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) Because: 1. Compiling without warnings helps assuring our users that the software we produce is of high quality (yes, clang warnings count). Yes, remove them and a gcc compiler warning occurs ... so, stalemate ;) 2. I don't want this ugly epidemic to spread. :) Ugly by Your definition... 3. I don't see you do the same in other pieces of code you write. Could be that I did not notice, was sleepy/lazy that day, etc, etc. Pick one. And last, but not least: the guidelines about following the surrounding code are about adding code, not re-factoring. Umm, wrong: That means that if you work on code that already exists Hmmm, I think this code already existed at the time of this change... -- Tom. Anyway, I think we have better things to do (like actually write code and have productive technical discussions) than to sit here and argue about thisbut, in case you don't, I will make some popcorn for the movie (Format Wars: Episode 9000). dh -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one
[E-devel] ecore_animator_pos_map issues
Hello, I wanted to write some tests for ecore_animator and while going through the functionality of ecore_animator_pos_map I noticed some inconsistencies. Here's how I *think* this function should behave: The docs say ecore_animator_pos_map() maps an input position from 0.0 to 1.0 along a timeline to a position in a different curve. It makes an input position (0.0 to 1.0) and maps to a new position (normally between 0.0 and 1.0, but it may go above/below 0.0 or 1.0 to show that it has overshot the mark) using some interpolation (mapping) algorithm. This means we can safely assume that the input is always between 0.0 and 1.0. Out-of-range values can be clamped to the min/max (and they are in the code). Now the return value of that function could be anything... It could even be less than 0.0 or more than 1.0. This is okay, but what is not explicitly stated (and which should be a requirement) is that pos_map(0.0) == 0.0 and pos_map(1.0) == 1.0 should hold true for any kind of transformation. We always want to start at the beginning and we always want to stop at the end... While writing the tests I found that using ECORE_POS_MAP_ACCELERATE pos_map(1.0) = 0.996982. Decelerate is wrong at 0.0 and sinusoidal also does not finish at 1.0. Linear is fine and I haven't tried the others yet. When moving 1000 pixels across the screen that error already amounts to 3 pixels so this should be significant. I traced that back to the eina_fp math (and maybe I should start by writing tests for that...). When using floating point functions from math.h the results were correct. I assume that eina_fp was/is used to improve speed on low-end machines? Does it still make sense to use that in this day and age as a default? Regards, Daniel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On Thursday 16 May 2013 17:37, Chris Michael wrote : On 16/05/13 17:29, Tom Hacohen wrote: On 16/05/13 17:17, Chris Michael wrote: On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) Because: 1. Compiling without warnings helps assuring our users that the software we produce is of high quality (yes, clang warnings count). that's what I was trying to reach, not for git blame lines, personal stats or preference (just saying), but for freeBSD.org, clang.debian.net, bitrig.org and the next to come… Yes, remove them and a gcc compiler warning occurs ... so, stalemate ;) I obviously checked before committing, gcc (GCC) 4.8.0 20130502 (prerelease) does not, I hear a voice from the future saying: remove this double parenthesis 2. I don't want this ugly epidemic to spread. :) Ugly by Your definition... 3. I don't see you do the same in other pieces of code you write. Could be that I did not notice, was sleepy/lazy that day, etc, etc. Pick one. And last, but not least: the guidelines about following the surrounding code are about adding code, not re-factoring. Umm, wrong: That means that if you work on code that already exists Hmmm, I think this code already existed at the time of this change... -- Tom. Anyway, I think we have better things to do (like
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On 16/05/13 19:08, Jérémy Zurcher wrote: On Thursday 16 May 2013 17:37, Chris Michael wrote : On 16/05/13 17:29, Tom Hacohen wrote: On 16/05/13 17:17, Chris Michael wrote: On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) Because: 1. Compiling without warnings helps assuring our users that the software we produce is of high quality (yes, clang warnings count). that's what I was trying to reach, not for git blame lines, personal stats or preference (just saying), but for freeBSD.org, clang.debian.net, bitrig.org and the next to come… Yes, remove them and a gcc compiler warning occurs ... so, stalemate ;) I obviously checked before committing, gcc (GCC) 4.8.0 20130502 (prerelease) does not, Great to hear ... but older versions do...I've seen it myself. dh I hear a voice from the future saying: remove this double parenthesis 2. I don't want this ugly epidemic to spread. :) Ugly by Your definition... 3. I don't see you do the same in other pieces of code you write. Could be that I did not notice, was sleepy/lazy that day, etc, etc. Pick one. And last, but not least: the guidelines about following the surrounding code are about adding code, not re-factoring. Umm, wrong: That means that if you work on code that already exists Hmmm, I think this code already existed at the time of this change... -- Tom. Anyway, I think we have better things to do (like actually write code and have
Re: [E-devel] [EGIT] [core/evas_generic_loaders] master 02/02: add glib-2 and librsvg version checks
On Wed, 15 May 2013 07:22:42 -0700 Jérémy Zurcher - Enlightenment Git no-re...@enlightenment.org said: what is the librsvg version check for including rsvg-cairo.h meant to be for? it doesn't work. are you trying to say that BEFORE 2.36.2 we HAD to include rsvg-cairo.h ourselves, but as of this version we no longer need to because rsvg automatically includes it? because... your check is broken, the ! in front does not.. work... jeyzu pushed a commit to branch master. commit 3912f569b0bf0771d4bcfc45e89c020aa5d10f44 Author: Jérémy Zurcher jer...@asynk.ch Date: Wed May 15 16:22:24 2013 +0200 add glib-2 and librsvg version checks --- src/bin/svg/main.c | 4 1 file changed, 4 insertions(+) diff --git a/src/bin/svg/main.c b/src/bin/svg/main.c index 366fe60..89a48fa 100644 --- a/src/bin/svg/main.c +++ b/src/bin/svg/main.c @@ -8,7 +8,9 @@ #include Eina.h #include librsvg/rsvg.h +#if !LIBRSVG_CHECK_VERSION(2,36,2) #include librsvg/rsvg-cairo.h +#endif #define DATA32 unsigned int @@ -57,7 +59,9 @@ static int _svg_init(const char *file) { #ifdef HAVE_SVG_2_36 +# if !defined(GLIB_VERSION_2_36) g_type_init(); +# endif #else rsvg_init(); #endif -- -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] A hell lot of Conditional jump or move depends on uninitialised value
On Thu, 16 May 2013 17:06:49 +0100 Tom Hacohen tom.haco...@samsung.com said: Yo, valgrind elementary_test List When starting: ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x58776FA: _evas_image_file_header (evas_image_load.c:191) ==20532==by 0x5877BCF: evas_common_load_rgba_image_module_from_file (evas_image_load.c:292) ==20532==by 0x583D9ED: _evas_cache_image_entry_new (evas_cache_image.c:279) ==20532==by 0x583F091: evas_cache_image_request (evas_cache_image.c:832) ==20532==by 0x587956B: evas_common_load_image_from_file (evas_image_main.c:766) ==20532==by 0x58A2FA3: eng_image_load (evas_engine.c:957) ==20532==by 0x57B9FA7: _image_file_set (evas_object_image.c:662) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) ==20532==by 0x64900D1: _eo_dov_internal (eo.c:400) ==20532==by 0x649030D: eo_do_internal (eo.c:429) ==20532==by 0x57B9D62: evas_object_image_file_set (evas_object_image.c:633) ==20532==by 0x4F54D1B: _elm_image_smart_file_set (elm_image.c:923) After scrolling a bit (all the way and back) and pressing the Next API function a few times (scroll all the way!) I get a storm of errors, for example: ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x57D0550: evas_object_render_pre_effect_updates (evas_object_main.c:470) ==20532==by 0x57C8839: evas_object_image_render_pre (evas_object_image.c:4375) ==20532==by 0x582C422: _evas_render_phase1_direct (evas_render.c:326) ==20532==by 0x5830840: evas_render_updates_internal (evas_render.c:1585) ==20532==by 0x58328BC: _canvas_render_async (evas_render.c:2067) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) ==20532==by 0x64900D1: _eo_dov_internal (eo.c:400) ==20532==by 0x649030D: eo_do_internal (eo.c:429) ==20532==by 0x5832823: evas_render_async (evas_render.c:2057) ==20532==by 0x141F8683: _ecore_evas_x_render (ecore_evas_x.c:661) ==20532==by 0x5D989F3: _ecore_evas_idle_enter (ecore_evas.c:98) ==20532==by 0x5B64964: _ecore_call_task_cb (ecore_private.h:303) ==20532== Conditional jump or move depends on uninitialised value(s) ==20532==at 0x582B663: evas_object_clip_recalc (evas_inline.x:294) ==20532==by 0x582C74D: _evas_render_phase1_object_process (evas_render.c:408) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582CBF0: _evas_render_phase1_object_process (evas_render.c:512) ==20532==by 0x582D047: _evas_render_phase1_process (evas_render.c:669) ==20532==by 0x5830714: evas_render_updates_internal (evas_render.c:1565) ==20532==by 0x58328BC: _canvas_render_async (evas_render.c:2067) ==20532==by 0x648FF6C: _eo_op_internal (eo.c:365) To me it smells like a non-potty trained CoW. But it might be something else. Completely up to date with everything (as of this moment). i shall begin with my first port of call when i see a bug. BLAME CEDRIC! :) -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] [EGIT] [core/efl] master 01/05: ecore evas wayland: if(()) are used for assignments, add missing NULL fields
On Thu, 16 May 2013 17:17:55 +0100 Chris Michael devilho...@comcast.net said: On 16/05/13 16:58, Tom Hacohen wrote: On 16/05/13 16:31, Rafael Antognolli wrote: On Thu, May 16, 2013 at 10:45 AM, Christopher Michael cp.mich...@samsung.com wrote: On 16/05/13 14:43, Daniel Juyung Seo wrote: On Thu, May 16, 2013 at 7:00 PM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 14:24, Christopher Michael wrote: On 16/05/13 14:19, Tom Hacohen wrote: On 16/05/13 14:17, Christopher Michael wrote: On 16/05/13 14:14, Tom Hacohen wrote: On 16/05/13 14:06, Christopher Michael wrote: On 16/05/13 14:06, Jérémy Zurcher wrote: sorry, but what are those formatting changes ?? Removing the parens that were there. … I see your commits, clang yells loud about this, many people are moving from gcc to clang, your code will keep yelling so, not really an issue for me. Really could care less what clang says to be honest :) They are ?? who ?? Distros still ship with gcc as the default compiler afaikWell, it does not yell here so not really an issue for me either ;) To be fair, that (()) clutter is ugly Ugly ?? Have you ever looked inside Elementary code ?? Now THAT's ugly ;) and should be removed even if clang doesn't complain. Why do you care so much for it anyway? -- Tom. It helps me keep my sanity when dealing with unruley if blocks and truth tests. Well, one man's ugly is another man's beauty I suppose ;) How does it help you? I'm genuinely interested. -- Tom. Order of precedence and readability mainly. Double parenthesis don't change the order of precedence. It's fine (and required by our conventions) if you had an AND or OR there, but since you don't have those, it just looks weird. +1. if ((ee-alpha == alpha)) return; (( === aaa)) looks weird to me. Yea, that looks weird to me too '===' ?? ;) It works but I am eager to clean this up. Anyhow, Jeremy could split the formatting fix commit and adding missing NULL commit. I think that was a point of devilhorns' mail. Well, that was one point, surebut my main point was... Don't change the formatting that was previously there please ;) Actually, wouldn't it be better if we try to follow EFL formatting inside the engine code? Of course I also do several mistakes regarding that formatting, but IMHO when this kind of discussion appears, we should just stick to the EFL formatting itself. We should stick to the formatting even before this kind of discussions appear. Yup Our coding guidelines don't really say anything regarding over resynthesizing, Exactly. It don't say anything about it. However, it does say: Our golden rule of coding - *FOLLOW THE CODING STYLE ALREADY THERE*. That means that if you work on code that already exists, keep to the spacing, indenting, variable and function naming style, etc. that already exists. short if (cond) action are fine as single line; use parenthesis for every clause or math; It also says: Use parenthesis to make clear what you want, even if the operator precedence is obvious to you. So yes, our standard does not say anything about the extra parens. but I'm quite certain, that if it had anything, it would have been a clear DO NOT DO. Especially in this kind of case where it doesn't and will never make any sense. Is that because you think it's ugly ?? Well, it makes sense to me...clearly defining the the condition. I don't understand why you are making such a big deal out of an extra pair of parens (that do not hurt or do anything, except maybe making readability better) in code that you don't maintain (and I doubt will ever even read) in this case the extra parens i personally see as fluff that your eye has to scan over ... and they add no extra meaning or intent. the rules are meant to encourage using ()'s to display intent. ie if (a == b c d || e ^ x 0x1f == 0x1) ... did i REALLY mean all of that as it reads with exact operator precedence rules? it's also just ... not a nice to read as: if ((a == b) (c d) || ((e ^ (x 0x1f)) == 0x1)) ... now i KNOW x 0x1f is meant to be done BEFORE the ^ operation. i know for sure. i also read the other operations/comparisons in their groups. if (x == y) and if ((x == y)) doesn't show me ANY more intention or meaning. it just makes the line longer with more visual noise to sift through. tom's right in that just commits fixing formatting tend to pollute git blame (or svn annotate). a few here or there i guess don't get hit very often, but if its done en-masse or in large bulk, then someone is more likely to be blamed when all it was was.. formatting fixes. so it's a trade-off. improve readability but hurt blaming... but blaming... that is ok. we all know to blame cedric. :) it's standard procedure. -- - Codito, ergo sum - I code, therefore I am
Re: [E-devel] Disk selector is in such a bad case it's not even funny
On Fri, May 17, 2013 at 12:27 AM, Tom Hacohen tom.haco...@samsung.com wrote: On 16/05/13 16:11, Tom Hacohen wrote: Maintainers of this stupid widget, please fix it. No maintainers atm. I found out that it was in a very bad state so fixing it was on my todo list. Thanks for the remind. Daniel Juyung Seo (SeoZ) Ways to reproduce: elementary_test Disk selector See how slow, stupid and full of errors it is. Bad *state. :) I'm also in a bad case, hm... state. -- Tom. -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- AlienVault Unified Security Management (USM) platform delivers complete security visibility with the essential security capabilities. Easily and efficiently configure, manage, and operate all of your security controls from a single console and one unified framework. Download a free trial. http://p.sf.net/sfu/alienvault_d2d ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel