Georg Baum wrote:
> Can you rerun valgrind with
>
> --track-origins=yes
>
> ? That should give us a hint where the initialization is missing in your
> case.
I uploaded it here:
http://www.lyx.org/trac/attachment/ticket/7371/valgrind2.log.bz2
Jürgen
Jürgen Spitzmüller wrote:
> Georg Baum wrote:
>
>> The attached patch
>> is a shot in the dark.
>
> Unfortunately, it does not help.
I suspected that, but there was a tiny chance...
>> I also notice a lot of uninitialised value warnings for
>> QTransform::fromTranslate() in your valgrind log.
Am 19.03.2011 um 14:18 schrieb Jürgen Spitzmüller:
> Richard Heck wrote:
FWIW that is also the version I have used in my tests.
>>>
>>> It might explain why this bug shows up only now (also in branch).
>>
>> FWIW, I'm using 4.7.1, and can't reproduce.
>
> Do you have the possibility to che
Richard Heck wrote:
> >> FWIW that is also the version I have used in my tests.
> >
> > It might explain why this bug shows up only now (also in branch).
>
> FWIW, I'm using 4.7.1, and can't reproduce.
Do you have the possibility to check with 4.7.2 as well?
Jürgen
On 03/19/2011 05:33 AM, Jürgen Spitzmüller wrote:
José Matos wrote:
Qt 4.7.2.
Jürgen
FWIW that is also the version I have used in my tests.
It might explain why this bug shows up only now (also in branch).
FWIW, I'm using 4.7.1, and can't reproduce.
rh
Jürgen Spitzmüller wrote:
> No, that's a safe way for me to get it (if I run an installed binary, that
> is).
>
> I also get it if I insert an note inset and change a color in preferences (so
> it's no genuine problem of the document dialog).
maybe another dataloss problem in dialog (#7349) is
Am 19.03.2011 um 10:33 schrieb Jürgen Spitzmüller:
> José Matos wrote:
>>> Qt 4.7.2.
>>>
>>>
>>>
>>> Jürgen
>>
>> FWIW that is also the version I have used in my tests.
>
> It might explain why this bug shows up only now (also in branch).
This and the fact it has the potential to crash your
José Matos wrote:
> > Qt 4.7.2.
> >
> >
> >
> > Jürgen
>
> FWIW that is also the version I have used in my tests.
It might explain why this bug shows up only now (also in branch).
Jürgen
On Saturday 19 March 2011 07:49:18 Jürgen Spitzmüller wrote:
> Qt 4.7.2.
>
> Jürgen
FWIW that is also the version I have used in my tests.
--
José Abílio
Georg Baum wrote:
> No. I was unable to reproduce the problem (as Pavel, with the same qt
> version 4.6.3). The strange thing is that the background colors are only
> used for LaTeX output, so changing them should not influence any GUI color
> (actually it should, but it is not implemented. Uwe
Richard Heck wrote:
> I still can't reproduce it here, unfortunately, or I'd try to help. I
> set the French locale, created a document with a shaded box, changed the
> color of the box, and applied, but nothing bad happened. Any other advice?
No, that's a safe way for me to get it (if I run an
Jürgen Spitzmüller wrote:
> Jürgen Spitzmüller wrote:
>> It's the box inset's background color. If I remove the BgColor
>> specification from the shaded box definition, the error disappears
>
> It is a general problem with background colors or the color cache. I get
> the same problem when chang
On 03/18/2011 06:44 AM, Jürgen Spitzmüller wrote:
Vincent van Ravesteijn wrote:
Can you step through the code in debug mode and find out when hell
breaks loose ?
I tried. It's really difficult to find out. It must be either somewhere in
applyView() or classChanged(). The first debug message I g
Jürgen Spitzmüller wrote:
> I try to run with valgrind as well, maybe I get something more interesting.
I ran valgrind again. The log file is here:
http://www.lyx.org/trac/attachment/ticket/7371/valgrind.log.bz2
Jürgen
Pavel Sanda wrote:
> Jürgen Spitzmüller wrote:
> > Maybe we have an invalid color cache and/or color while redrawing the
> > screen? Does this ring a bell to someone?
>
> is it regression so we can bisect?
Actually, no. I can reproduce it in branch as well (changing the shaded
background prefs
Jürgen Spitzmüller wrote:
> Maybe we have an invalid color cache and/or color while redrawing the screen?
> Does this ring a bell to someone?
is it regression so we can bisect?
i have seen bformat in backtrace - shot in the dark - if you
remerge all strings, does it make any difference?
pavel
Jürgen Spitzmüller wrote:
> It's the box inset's background color. If I remove the BgColor
> specification from the shaded box definition, the error disappears
It is a general problem with background colors or the color cache. I get the
same problem when changing any color in document or prefs w
Jürgen Spitzmüller wrote:
> Vincent van Ravesteijn wrote:
> > Can you step through the code in debug mode and find out when hell
> > breaks loose ?
>
> I tried. It's really difficult to find out. It must be either somewhere in
> applyView() or classChanged(). The first debug message I get after th
Vincent van Ravesteijn wrote:
> Can you step through the code in debug mode and find out when hell
> breaks loose ?
I tried. It's really difficult to find out. It must be either somewhere in
applyView() or classChanged(). The first debug message I get after the weird
characters are output for t
Op 18-3-2011 11:32, Jürgen Spitzmüller schreef:
Jürgen Spitzmüller wrote:
I try to run with valgrind as well, maybe I get something more interesting.
Nothing that seems relevant. I tried some debugging, but found nothing so far.
Any ideas where I should look at?
Jürgen
Can you step through
Jürgen Spitzmüller wrote:
> I try to run with valgrind as well, maybe I get something more interesting.
Nothing that seems relevant. I tried some debugging, but found nothing so far.
Any ideas where I should look at?
Jürgen
Richard Heck wrote:
> It looks like there is a boolean there that hasn't been initialized. I
> suppose that could be causing us to read something bad. I've committed a
> repair at r37945.
Alas, this doesn't fix the problem.
Meanwhile I also noted that I can trigger the problem when changing a
On 03/17/2011 06:54 PM, Jean-Marc Lasgouttes wrote:
Le 17/03/11 19:57, José Matos a écrit :
I sent Jean-Marc the file 4 MB uncompressed and 180 KB compressed. If
anyone
else wants the valgrind report I can send it.
The problem may be related to the reportss below (the asserts begin
just afte
Le 17/03/11 19:57, José Matos a écrit :
I sent Jean-Marc the file 4 MB uncompressed and 180 KB compressed. If anyone
else wants the valgrind report I can send it.
The problem may be related to the reportss below (the asserts begin just
after them). I do not know what they mean, though.
This
On Thursday 17 March 2011 16:24:55 Jean-Marc Lasgouttes wrote:
> Le 17/03/2011 16:30, José Matos a écrit :
> > The message that should be there (at least compared with the case where
> > the language is English) is
> > "Converting document to new document documents class"
> >
> > But as I have tol
Le 17/03/2011 16:30, José Matos a écrit :
The message that should be there (at least compared with the case where the
language is English) is
"Converting document to new document documents class"
But as I have told before something is wrong here since it takes up to 5
seconds to apply the change
On Thursday 17 March 2011 15:16:10 Vincent van Ravesteijn wrote:
> Do you also see a connection with the shaded box. Any idea, what text
> should have been in the status bar at that moment ?
>
> Vincent
If you change first the document settings and only later insert the shaded box
then no probl
Op 17-3-2011 16:12, José Matos schreef:
On Wednesday 16 March 2011 17:24:52 Jürgen Spitzmüller wrote:
... and only if I run an installed LyX. If I run the lyx binary from the
build tree, the error neither triggers.
Jürgen
Here i happens if I start from the build directory, and just as you not
Am 16.03.2011 18:19, schrieb Jürgen Spitzmüller:
It seems the error only occurs if I use a localization other than
English. With LANG=en_EN, the error does not trigger, but I get it both
with LANG=de_DE and LANG=fr_FR.
... and only if I run an installed LyX. If I run the lyx binary from the
b
Am 16.03.2011 18:00, schrieb Jürgen Spitzmüller:
It looks like the message strings and/or the file name gets corrupted.
Look at the file name in the window title of the attached screen shot.
It seems the error only occurs if I use a localization other than
English. With LANG=en_EN, the error d
On Wednesday 16 March 2011 16:55:08 Richard Heck wrote:
> Nor here (Fedora 13).
>
> rh
Nor here, tested both on Fedora 14 and Fedora 15 (to be).
With that said there is something wrong going on since it says after pressing
OK that it says "Converting document to new document documents class" in
Am 16.03.2011 16:41, schrieb Stephan Witt:
Sorry, I cannot reproduce it on Mac with SVN.
It looks like the message strings and/or the file name gets corrupted.
Look at the file name in the window title of the attached screen shot.
Eventually LyX asserts with
lassert.cpp(21): ASSERTION conta
On 03/16/2011 11:41 AM, Stephan Witt wrote:
Am 16.03.2011 um 15:10 schrieb Jürgen Spitzmüller:
I cannot (yet) nail down this, and it's not always reproducible (actually, I
fail to reproduce it ATM), but I report it nevertheless since it strikes me to
be a severe problem (a release blocker):
Am 16.03.2011 um 15:10 schrieb Jürgen Spitzmüller:
> I cannot (yet) nail down this, and it's not always reproducible (actually, I
> fail to reproduce it ATM), but I report it nevertheless since it strikes me
> to be a severe problem (a release blocker):
>
> * insert a shaded box
> * go to docum
34 matches
Mail list logo
