Re: expose is broken
[EMAIL PROTECTED] (Lars Gullik Bjønnes) writes: | One of the last commits broke redraw of the WorkArea. > | Just load a document, open an menu, close it and see for yourselves. > | If not fixed fairly quickly the patch(es) must be reverted. It seems to me that the screen::expose is using the pixmaps wrong or something. even a "screen-recenter" does not work: Workarea event: KEYBOARD screen: drawFromTo 414-414 expose 667x0+0+414 This changes nothing on screen. Later if I move the cursor down until a scroll is done, the correct expose is run: Workarea event: KEYBOARD screen: drawFromTo 0-414 expose 667x414+0+0 I can now even trigger a "Total Grey Out", by moving _into_ the the workarea... - have the workarea partially covered by a couple of other windows - move focus to the workarea - Ta ta: Instant grey out. XForms expose/draw/redraw is now close to unusable, I am not qute sure when this happened or what caused it, but it must be fixed. Please to not do any futher cleanup in this area until this has been resolved. Direct your efforts to discover and this this instead. -- Lgb
Re: expose is broken
On Fri, Jun 21, 2002 at 11:33:53PM +0200, Lars Gullik Bjønnes wrote: > One of the last commits broke redraw of the WorkArea. > > Just load a document, open an menu, close it and see for yourselves. Don't see a problem. I see some problems, but not this one. > If not fixed fairly quickly the patch(es) must be reverted. Well this is a new one on me ! Please explain your new policy. And explain why this reversion policy did not apply to the disastrous changes pre-1.2.0 which took many many months to fix. I'm going away soon, so if you want to revert (planning a 1.3.0 tomorrow ?), you'll have to do it yourself. john -- "If a thing is not diminished by being shared, it is not rightly owned if it is only owned & not shared." - St. Augustine
Re: expose is broken
On Sat, Jun 22, 2002 at 01:48:27AM +0200, Lars Gullik Bjønnes wrote: > It seems to me that the screen::expose is using the pixmaps wrong or > something. More informed bug reports are welcome. > I can now even trigger a "Total Grey Out", by moving _into_ the the > workarea... >- have the workarea partially covered by a couple of other windows >- move focus to the workarea >- Ta ta: Instant grey out. If you want it fixed, you have to provide me with a reproducable testcase. And no, this is not it. At the very least a -dbg gui,workarea output. > XForms expose/draw/redraw is now close to unusable Well maybe I am drunk but this makes me fume. JMarc can't even use 1.3.0 AT ALL thanks to your broken changes and you're whining about some minor draw problems you could fix yourself if you took ten minutes. Revert your boost patches or STFU. Angus, perhaps you can actually provide some useful help with these problems, as you usually do. john -- "If a thing is not diminished by being shared, it is not rightly owned if it is only owned & not shared." - St. Augustine
Re: expose is broken
Herbert Voss <[EMAIL PROTECTED]> writes: | Lars Gullik Bjønnes wrote: > >> One of the last commits broke redraw of the WorkArea. >> Just load a document, open an menu, close it and see for yourselves. > > | just like http://www.lyx.org/~voss/workarea.gif exactly. >> If not fixed fairly quickly the patch(es) must be reverted. > > | as John said ... I pay exaclty no attention to mumbling. -- Lgb
Re: expose is broken
John Levon <[EMAIL PROTECTED]> writes: | On Sat, Jun 22, 2002 at 01:48:27AM +0200, Lars Gullik Bjønnes wrote: > >> It seems to me that the screen::expose is using the pixmaps wrong or >> something. > | More informed bug reports are welcome. Meaning that my report is wrong or that you want more details? >> I can now even trigger a "Total Grey Out", by moving _into_ the the >> workarea... >>- have the workarea partially covered by a couple of other windows >>- move focus to the workarea >>- Ta ta: Instant grey out. > | If you want it fixed, you have to provide me with a reproducable | testcase. And no, this is not it. At the very least a -dbg gui,workarea | output. You got parts of it. >> XForms expose/draw/redraw is now close to unusable > | Well maybe I am drunk but this makes me fume. JMarc can't even use 1.3.0 | AT ALL thanks to your broken changes and you're whining about some minor | draw problems you could fix yourself if you took ten minutes. I took ten minutes and could not. And I am at least pestering someone about the problems I see. If you (or anyone else) just keep quiet how am I supposed to know that you still have a problem. And I do not call the failure to redraw any part of the screen on an expose a minor problem. | Revert your boost patches or STFU. Because the boost changes are making a _running_ lyx misbehave in what way? (and are you really unable to keep issues separate?) -- Lgb
Re: expose is broken
Lars Gullik Bjønnes wrote: > One of the last commits broke redraw of the WorkArea. > > Just load a document, open an menu, close it and see for yourselves. just like http://www.lyx.org/~voss/workarea.gif > > If not fixed fairly quickly the patch(es) must be reverted. as John said ... Herbert -- http://www.lyx.org/help/
Re: expose is broken
On Sat, Jun 22, 2002 at 08:42:19AM +0200, Lars Gullik Bjønnes wrote: > | More informed bug reports are welcome. > > Meaning that my report is wrong or that you want more details? More details. But looking it's obvious - your WM doesn't do saveunders for menus, and we're not doing the copy of the pixmap at FL_DRAW time. Related to the stuff me & angus were talking aobut. > > And I am at least pestering someone about the problems I see. > If you (or anyone else) just keep quiet how am I supposed to know that > you still have a problem. Yes, I remember how quiet everyone was about the boost changes. > Because the boost changes are making a _running_ lyx misbehave in what > way? (and are you really unable to keep issues separate?) Because I was drunk gimme a break ... regards john -- "If a thing is not diminished by being shared, it is not rightly owned if it is only owned & not shared." - St. Augustine
Re: expose is broken
On Monday 24 June 2002 12:02 pm, John Levon wrote: > On Sat, Jun 22, 2002 at 08:42:19AM +0200, Lars Gullik Bjønnes wrote: > > | More informed bug reports are welcome. > > > > Meaning that my report is wrong or that you want more details? > > More details. But looking it's obvious - your WM doesn't do saveunders > for menus, and we're not doing the copy of the pixmap at FL_DRAW time. > Related to the stuff me & angus were talking aobut. As I see it, the current code is much the same as the old code. The philosophy behind the code has not changed yet anyway. So do the changes you've introduced so far fall between two stools? Ie, they are neither the new way nor the old way, they're just broken? More importantly, is it fixable without going the whole way? As you'll gather, I'm confused by what you actually did commit on Friday. Could you post the patch and/or explain it? Certainly give an explantation of what you did and why. Help us keep up with you ;-) > > And I am at least pestering someone about the problems I see. > > If you (or anyone else) just keep quiet how am I supposed to know that > > you still have a problem. > > Yes, I remember how quiet everyone was about the boost changes. > > > Because the boost changes are making a _running_ lyx misbehave in what > > way? (and are you really unable to keep issues separate?) > > Because I was drunk gimme a break ... Bad behaviour when drunk is still bad behaviour. STFU is no way to behave. An old prude.
Re: expose is broken
John Levon <[EMAIL PROTECTED]> writes: | On Sat, Jun 22, 2002 at 08:42:19AM +0200, Lars Gullik Bjønnes wrote: > >> | More informed bug reports are welcome. >> >> Meaning that my report is wrong or that you want more details? > | More details. But looking it's obvious - your WM doesn't do saveunders | for menus, and we're not doing the copy of the pixmap at FL_DRAW time. | Related to the stuff me & angus were talking aobut. _but_ I have never had problems with LyX and saveunders with this WM ealier, so it must be somehting that we do different. >> And I am at least pestering someone about the problems I see. >> If you (or anyone else) just keep quiet how am I supposed to know that >> you still have a problem. > | Yes, I remember how quiet everyone was about the boost changes. I have tried to act on some of those issues, esp. binary size. For gcc 3.0 the binary size is imo more than satisfactory, for 2.96 it as imho acceptable albeit a bit big. >> Because the boost changes are making a _running_ lyx misbehave in what >> way? (and are you really unable to keep issues separate?) > | Because I was drunk gimme a break ... drunkards get no leeway. -- Lgb
Re: expose is broken
> "John" == John Levon <[EMAIL PROTECTED]> writes: John> Well maybe I am drunk but this makes me fume. JMarc can't even John> use 1.3.0 AT ALL thanks to your broken changes and you're John> whining about some minor draw problems you could fix yourself if John> you took ten minutes. Concerning the fact that I cannot run 1.3.0, I have not been complaining too much about it recently because I hardly have enough time to work on 1.2.1 these days. Of course, this means that every patch I apply to 1.2.1cvs is also applied to 1.3.0cvs, but blindly (I only check that it compiles). This is certainly annoying, to say the least. I will have some free time in a couple of weeks and will try to investigate seriously then. I have to admit that I thought this issue would be discussed/solved in porto, and I am a bit disappointed to see that it seem it was not even discussed there. JMarc
Re: expose is broken
On Monday 01 July 2002 12:31, Jean-Marc Lasgouttes wrote: > > "John" == John Levon <[EMAIL PROTECTED]> writes: > > Concerning the fact that I cannot run 1.3.0, I have not been > complaining too much about it recently because I hardly have enough > time to work on 1.2.1 these days. Of course, this means that every > patch I apply to 1.2.1cvs is also applied to 1.3.0cvs, but blindly (I > only check that it compiles). This is certainly annoying, to say the > least. I will have some free time in a couple of weeks and will try to > investigate seriously then. That I know that it doesn't help, but after John last patch I am unable to run LyX at all. It core dumps at start with a segfault. So I think that I can complain also. Before that, 1.3.0cvs always segfaulted at end due to an unfulfilled assert. > I have to admit that I thought this issue would be discussed/solved in > porto, and I am a bit disappointed to see that it seem it was not even > discussed there. Actually this was discussed, because I took to the meeting an Imac running mandrake 8.2 that had the same problem as you do. Lars concluded that it should be something related with gcc 2.95.3. Am I right here, Lars? > JMarc -- José Abílio
Re: expose is broken
> "José" == José Abílio Oliveira Matos <[EMAIL PROTECTED]> writes: >> I have to admit that I thought this issue would be discussed/solved >> in porto, and I am a bit disappointed to see that it seem it was >> not even discussed there. José> Actually this was discussed, because I took to the meeting an José> Imac running mandrake 8.2 that had the same problem as you do. José> Lars concluded that it should be something related with gcc José> 2.95.3. Am I right here, Lars? I guess you mean 2.96. Did you actually try to debug the problem? I'd appreciate at least a workaround, since I think lyx 1.3.0 will have to work on mdk 8.[12] when it is relelased. JMarc
Re: expose is broken
On Mon, Jul 01, 2002 at 01:31:14PM +0200, Jean-Marc Lasgouttes wrote: > I have to admit that I thought this issue would be discussed/solved in > porto, and I am a bit disappointed to see that it seem it was not even > discussed there. I have the impressions that things are a bit better know although I don't know whether there was some actual change or whether I just got used to it. What was your problem exactly? [Pointer to archive would suffice] Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
Re: expose is broken
On Mon, Jul 01, 2002 at 12:42:00PM +0100, José Abílio Oliveira Matos wrote: > That I know that it doesn't help, but after John last patch I am unable to > run LyX at all. It core dumps at start with a segfault. So I think that I can > complain also. Things are only slightly better here: It runs, but it is unsuable because the screen turns grey all the time. Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
Re: expose is broken
> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: Andre> On Mon, Jul 01, 2002 at 01:31:14PM +0200, Jean-Marc Lasgouttes Andre> wrote: >> I have to admit that I thought this issue would be discussed/solved >> in porto, and I am a bit disappointed to see that it seem it was >> not even discussed there. Andre> I have the impressions that things are a bit better know Andre> although I don't know whether there was some actual change or Andre> whether I just got used to it. It still does not start here. Andre> What was your problem exactly? [Pointer to archive would Andre> suffice] There is a valgrind backtrace here http://www.mail-archive.com/lyx-devel@lists.lyx.org/msg39397.html Problem is that I cannot even launch lyx. It crashes when setting up its slots in bufferview_pimpl constructor. A double-free problem. JMarc
Re: expose is broken
Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: >> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes: > | Andre> On Mon, Jul 01, 2002 at 01:31:14PM +0200, Jean-Marc Lasgouttes | Andre> wrote: >>> I have to admit that I thought this issue would be discussed/solved >>> in porto, and I am a bit disappointed to see that it seem it was >>> not even discussed there. > | Andre> I have the impressions that things are a bit better know | Andre> although I don't know whether there was some actual change or | Andre> whether I just got used to it. > | It still does not start here. > | Andre> What was your problem exactly? [Pointer to archive would | Andre> suffice] > | There is a valgrind backtrace here | http://www.mail-archive.com/lyx-devel@lists.lyx.org/msg39397.html > | Problem is that I cannot even launch lyx. It crashes when setting up | its slots in bufferview_pimpl constructor. A double-free problem. Have you tried to compile with -O0? (and -fno-inline) -- Lgb
Re: expose is broken
On Mon, Jun 24, 2002 at 02:34:30PM +0200, Lars Gullik Bjønnes wrote: > _but_ I have never had problems with LyX and saveunders with this WM > ealier, so it must be somehting that we do different. readd an unconditional updateScreen() call to BufferView_pimpl.C:workAreaResize(). This should fix xforms. I would check if my box wasn't packed away, sorry I can't help more right now ... regards john -- "If a thing is not diminished by being shared, it is not rightly owned if it is only owned & not shared." - St. Augustine
Re: expose is broken
On Tue, Jul 02, 2002 at 04:07:35PM +0100, John Levon wrote: > > _but_ I have never had problems with LyX and saveunders with this WM > > ealier, so it must be somehting that we do different. > > readd an unconditional updateScreen() call to > BufferView_pimpl.C:workAreaResize(). And how does this translate into English? Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
Re: expose is broken
On Tue, Jul 02, 2002 at 05:10:37PM +0200, Andre Poenitz wrote: > > readd an unconditional updateScreen() call to > > BufferView_pimpl.C:workAreaResize(). > > And how does this translate into English? dunno how to make it clearer ? Perhaps : vi BufferView_pimpl.C /::workAreaResize /^} i updateScreen(); :wq make As I said I am unable to test this, so it's pulled out of my arse. Please add a "// FIXME: GUII this is wrong" above it too. I wouldn't have made the error, but my WM uses SaveUnders (I guess), so it was only noticable with fvwm etc. Sorry ! What used to happen was that xforms on getting an expose event, would send a workAreaExpose(), which would call screen.redraw() (I think). This would then realise that the view hasn't scrolled, and nothing needs to be redrawn, so the whole thing is just copied to the visible window from the backing pixmap. I'd changed it to work the sensible way, but had accidentally missed the case where the window gets an expose event, but it has not been generated by lyx itself, and the window has not changed geometry. Hmm, you might need to always fire a workAreaResize() too from XWorkArea.C (search for workAreaResize()). Please experiment. Sorry I can't be more help right now. regards john -- "If a thing is not diminished by being shared, it is not rightly owned if it is only owned & not shared." - St. Augustine
Re: expose is broken
John Levon <[EMAIL PROTECTED]> writes: | On Tue, Jul 02, 2002 at 05:10:37PM +0200, Andre Poenitz wrote: > >> > readd an unconditional updateScreen() call to >> > BufferView_pimpl.C:workAreaResize(). >> >> And how does this translate into English? > | dunno how to make it clearer ? Perhaps : > | vi BufferView_pimpl.C | /::workAreaResize | /^} | i updateScreen(); | :wq | make > | As I said I am unable to test this, so it's pulled out of my arse. | Please add a "// FIXME: GUII this is wrong" above it too. > | I wouldn't have made the error, but my WM uses SaveUnders (I guess), so | it was only noticable with fvwm etc. Sorry ! The error is there with kwm as well. -- Lgb