Re: Glib for iOS
On 07/18/2012 12:49 AM, Hubert Figuière wrote: On 17/07/12 03:46 PM, Paul Davis wrote: it can be very hard. there are two approaches. one is to build a pseudo/fake gtk-doc package for OS X, which does actually exist, but i can't tell you where to find it. google will help. the other is to install ALL the dependencies that gtk-doc requires, which is substantial: And do we know the rationale for forcing gtk-doc ? I mean glib is supposed to be lightweight library, but this is like everything including the kitchen sink. This alone as always discouraged me to use glib for portable code. This only happens if you build from e.g. git. It does not happen if you build from a tarball. The issue is that you need gtk-doc.m4 in order to resolve the macro, therwise the macro invocation stays there in the generated configure file, which then won't run. There are efforts to resolv it: https://bugzilla.gnome.org/show_bug.cgi?id=674314 Stefan Hub ___ gtk-devel-list mailing list gtk-devel-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-devel-list ___ gtk-devel-list mailing list gtk-devel-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-devel-list
About making public gtk accessibility headers
The conclusion of a recent thread was discuss gtk technical stuff on this list, so I will do that instead of creating a bug. Background: at the end of the GUADEC we had a BOF/Camp about accessibility [1]. During one of those days we were taking a look to evince. Although evince accessibility support is still a work in progress, some months there was something working, and now nothing is working. The reason is some of the lately changes on gtk3 in relation of how to specify the accessibility class to a specific gtk class. In the old times the way to specify it was using factories [4]. Factories were used to set the type of the accessibility object, and also to get the type of the parent accessible object. Something that (perhaps) made sense on a environment were the ATK implementation was in a plugin. These days that is not true anymore. ATK implementation for GTK are included on GTK. After the last changes on gtk, gtkwidget saves the accessible_type to use on each class, so in order to specify what it is the type that you want to use is as easy as call gtk_widget_class_set_accessible_type at the class_init. So, what is doing gtk_widget_get_accessible?: * If you have a valid accessible type at the widget class, use it to instance the accessible object. * If the accessible type at the widget class is GTK_TYPE_ACCESSIBLE, it tries to use the factories. The second one is something like a fallback. But that fallback is failing. The example is evince. They have a custom widget, EvView, subclass of GtkLayout. Until now evince was using the factories to specify EvViewAccessible as the type to be used to instance EvView accessible object, and also in order to get the type of the accessible object assigned to GtkLayout. But the accessible type is already is already assigned (at GtkContainer class_init). So the factories are not being used on EvView-get_accessible. So, the obvious solution could be use gtk_widget_class_set_accessible_type on EvView. But the problem is that the headers for the accessibility objects on gtk are not public (in fact, if you take a look to gtk/a11y get_type methods has a '_' prefix so we assume that they are private). Probably we could keep doing some kind of hack to obtain the parent accessible type. But remember that we are talking about hacks like the ones on link [2] and specially link [3]. IMHO, the best solution would be making public also the accessibility headers, so any third-party doing custom widgets could use directly the types of those classes to extend, and just one line to specify the accessibility type. Take into account that evince are not the only one affected. Any application defining custom widgets are also affected. I also don't see the API of those headers changing a lot, as most of the functionality is defined on ATK interfaces (although something like adding pads on those could be sensible). The only possible problem could arise if we decide to add an accessible class in the middle of two existing classes (like a new gtkbinaccessible in the middle of current gtkcontaineraccessible and gtkbuttonaccessible), as the parent class could change. But I see that unlikely at this moment. Finally, take into account that clutter is already doing that. Cally (clutter ATK implementation, that still uses factories, but eventually they will be nuked) is public, so St at gnome-shell is free-factory. Opinions? BR [1] https://live.gnome.org/Accessibility/BoFs/A11yCamp2012 [2] http://projects.gnome.org/accessibility/talks/GUAD3C/making-apps-accessible/anonymous-inheritance-1.html [3] http://projects.gnome.org/accessibility/talks/GUAD3C/making-apps-accessible/anonymous-inheritance-2.html [4] http://developer.gnome.org/atk/stable/AtkObjectFactory.html -- Alejandro Piñeiro Iglesias ___ gtk-devel-list mailing list gtk-devel-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-devel-list
Re: GTK+ meetup at Guadec
On Sun, Jul 29, 2012 at 09:22:51AM +1000, Andrew Cowie wrote: On Sat, 2012-07-28 at 15:30 +0200, Matthias Clasen wrote: - We want to discuss major API proposals or wishes (eg new widgets) on this mailing list, instead of just in bugzilla. That's terrific. While I wouldn't say that bugzilla not in the public record or anything like that, the fact that it is blocked from search indexes makes it hard to stay abreast of relevant discussion even if you _are_ subscribed to bugmail. Convince chpe that it is ok to have his comments indexed and I'll gladly have Bugzilla indexed. -- Regards, Olav ___ gtk-devel-list mailing list gtk-devel-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-devel-list
Re: Using Windows, Eclipse, Python - gtk not importing
On Tue, 31 Jul 2012 16:31:41 -0700, Mark Tiddens wrote: So, maybe I need to do something to get Eclipse or Python to find gtk? - In Eclipse's menu, go to Window Preferences PyDev Interpreter - Python - In the Python interpreters list, select the Python interpreter you want to configure (this example assumes Python 2.7 installed into C:\bin\Python27, adjust as needed) - In the bottom notebook, go to the Environment tab and click the New... button - Set Name: to PATH and Value: to C:\bin\Python27\Lib\site-packages\gtk-2.0\runtime\bin;%PATH% and click the OK button. The Value: part points to the bin directory of your GTK+ bundle (this example assumes the PyGTK All-in-one installer which already comes with an appropriate GTK+ bundle out of the box). - In the bottom notebook, go to the Forced Builtins tab and click the New... button - Set Builtin to add to cairo,glib,gobject,gio,atk,pango,pangocairo,gtk,gtk.gdk,rsvg,gtksourceview2,goocanvas and click the OK button. - Back in the Preferences dialog, click the OK button and wait for PyDev to update its symbol database. So, the first part (setting PATH) answers your question, the second part (setting forced builtins) ensures PyDev's more advanced features like code completion function correctly for PyGTK's C extension modules and works for Eclipse on Windows and Linux equally well. Note: sometimes PyDev fails to notice its symbol database has been updated and various combinations of restarting Eclipse, right-clicking your PyDev project and selecting PyDev Remove error markers and PyDev Code analysis are needed before the Python editor stops showing Unresolved import errors etc... Through the years this seems to have happened for some combinations of the Eclipse platform and PyDev version. It always eventually starts working though and once it does, PyDev is a joy to use. Especially its debugger, which has nothing even remotely competing with it imho :) mvg, Dieter ___ gtk-list mailing list gtk-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-list
Position of window-popup in GTK 3
one canfix the GtkToolbar-internal-padding: 0 on a TOOLBAR .. is I have trouble placing a window below a TOOLBUTTON ..in GTK 3. .I whenchange the theme in GNOME 3 the space in theTOOLBAR also changes . I may give an example ( INTERNET or CODE ).. best regards MARIANO ___ gtk-list mailing list gtk-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-list
Re: Hello, some of us still care about memory usage
On 07/31/2012 06:56 PM, Olav Vitters wrote: On Wed, Aug 01, 2012 at 12:35:07AM +0200, Olav Vitters wrote: I heard that your patch was really good. Cannot comment myself as I'm not a developer. Seems you've contributed a lot of good patches already. I've changed your Bugzilla permissions and I've pinged a few people about your bugs. Thanks. What do the new permissions allow? I noticed one of your bugs was about Windows. There is an entire mailing list for that too. Anyones patches should be looked at without any intervention being needed. But if needed, please cc me on bugs you think are being ignored (use bugzilla-gn...@vitters.nl, it works only for Bugzilla). Will do. -- John ___ gtk-list mailing list gtk-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-list
Re: Hello, some of us still care about memory usage
On Wed, Aug 01, 2012 at 09:37:09PM -0400, John Lindgren wrote: On 07/31/2012 06:56 PM, Olav Vitters wrote: On Wed, Aug 01, 2012 at 12:35:07AM +0200, Olav Vitters wrote: I heard that your patch was really good. Cannot comment myself as I'm not a developer. Seems you've contributed a lot of good patches already. I've changed your Bugzilla permissions and I've pinged a few people about your bugs. Thanks. What do the new permissions allow? Change any bug. Doesn't matter if you're the reporter or not. -- Regards, Olav ___ gtk-list mailing list gtk-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-list