I found that an access to a NULL-pointer caused the crashes on pdf/ps export.
The pointer access is in the file src/extension/prefdialog.cpp line 178 in the function: void PrefDialog::on_response (int signal) {
The function was last edited by gouldtj as follow: r17821@...1818...: ted | 2008-01-31 12:59:57 -0800 Okay, I'm not 100% sure what all these changes are but I want to start working on this again from a checked in point. I'll fix the bugs from here.
So if one can review the diff from 17244 to 16419. Right now I just test for NULL in that specific case I found.
HTH,
Adib.
the Adib wrote:
I found that an access to a NULL-pointer caused the crashes on pdf/ps export.
The pointer access is in the file src/extension/prefdialog.cpp line 178 in the function: void PrefDialog::on_response (int signal) {
The function was last edited by gouldtj as follow: r17821@...1818...: ted | 2008-01-31 12:59:57 -0800 Okay, I'm not 100% sure what all these changes are but I want to start working on this again from a checked in point. I'll fix the bugs from here.
So if one can review the diff from 17244 to 16419. Right now I just test for NULL in that specific case I found.
In fact, according to git-bisect the bug was introduced in that particular commit #17244. Also see
https://bugs.launchpad.net/inkscape/+bug/169066.
Would it make sense to roll back the changes of #17244 until this is is sorted out? Or does anyone have a clue what precisely is causing the crash so that it can be fixed?
Max
Ted, could you pls check what is the potential risk (and/or benefit) of your commit. And if we need to postpone your patch until after 0.46.
HTH,
Adib.
On Wed, Feb 20, 2008 at 11:46 PM, Maximilian Albert <Anhalter42@...173...> wrote:
the Adib wrote:
I found that an access to a NULL-pointer caused the crashes on pdf/ps export.
The pointer access is in the file src/extension/prefdialog.cpp line 178 in the function: void PrefDialog::on_response (int signal) {
The function was last edited by gouldtj as follow: r17821@...1818...: ted | 2008-01-31 12:59:57 -0800 Okay, I'm not 100% sure what all these changes are but I want to start working on this again from a checked in point. I'll fix the bugs from here.
So if one can review the diff from 17244 to 16419. Right now I just test for NULL in that specific case I found.
In fact, according to git-bisect the bug was introduced in that particular commit #17244. Also see
https://bugs.launchpad.net/inkscape/+bug/169066.
Would it make sense to roll back the changes of #17244 until this is is sorted out? Or does anyone have a clue what precisely is causing the crash so that it can be fixed?
Max
Maximilian Albert schrieb:
the Adib wrote:
I found that an access to a NULL-pointer caused the crashes on pdf/ps export.
Sorry, I didn't notice that you had already committed a fix to SVN. I could only give it a quick try but the crash is gone for me and ps/pdf export seems to work alright. What do others say?
Max
In nightly build for windows "Inkscape0802201629.7z" from http://inkscape.modevia.com/win32/?M=D PDF export doesn't work. But in nightly build "inkscape-0802201700.7z" from http://inkscape.modevia.com/win32-snap/?C=M;O=D it works again.
However, I'm on Windows XP, and export blurs as bitmaps has never worked so far. And if you uncheck "export text as paths", the text is still turned upside down.
On Thu, Feb 21, 2008 at 12:21 AM, Maximilian Albert <Anhalter42@...173...> wrote:
Maximilian Albert schrieb:
the Adib wrote:
I found that an access to a NULL-pointer caused the crashes on pdf/ps
export.
Sorry, I didn't notice that you had already committed a fix to SVN. I could only give it a quick try but the crash is gone for me and ps/pdf export seems to work alright. What do others say?
Max
This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Inkscape-devel mailing list Inkscape-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-devel
Heh, I just fixed this aswell last evening, but didn't have internet. I had the exact same fix :-)
Johan
-----Original Message----- From: inkscape-devel-bounces@lists.sourceforge.net [mailto:inkscape-devel-bounces@lists.sourceforge.net] On Behalf Of Maximilian Albert Sent: donderdag 21 februari 2008 0:22 To: the Adib Cc: Inkscape Devel List Subject: Re: [Inkscape-devel] crash on pdf/ps export
Maximilian Albert schrieb:
the Adib wrote:
I found that an access to a NULL-pointer caused the
crashes on pdf/ps export.
Sorry, I didn't notice that you had already committed a fix to SVN. I could only give it a quick try but the crash is gone for me and ps/pdf export seems to work alright. What do others say?
Max
This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Inkscape-devel mailing list Inkscape-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-devel
J.B.C.Engelen@...1578... schrieb:
Heh, I just fixed this aswell last evening, but didn't have internet. I had the exact same fix :-)
Sometimes I wonder whether NULL pointer tests really solve these issues in all cases or if they sometimes just cover other underlying bugs (maybe the very fact that a NULL pointer is passed is wrong?). But in this case the fix seems to work fine; at least my exported pdf/ps files look good. :)
However, the crash unfortunately still occurs when one presses "Close" instead of "OK" in the pdf/ps options dialog which appears after the "Save file" dialog.
Max
On Fri, Feb 22, 2008 at 01:36:19AM +0100, Maximilian Albert wrote:
J.B.C.Engelen@...1578... schrieb:
Heh, I just fixed this aswell last evening, but didn't have internet. I had the exact same fix :-)
Sometimes I wonder whether NULL pointer tests really solve these issues in all cases or if they sometimes just cover other underlying bugs (maybe the very fact that a NULL pointer is passed is wrong?). But in this case the fix seems to work fine; at least my exported pdf/ps files look good. :)
However, the crash unfortunately still occurs when one presses "Close" instead of "OK" in the pdf/ps options dialog which appears after the "Save file" dialog.
It's fairly traditional to use a NULL pointer to mean "not defined", so NULL pointer checks are quite standard and commonly used for controlling when certian logic should be invoked.
Bryce
-----Original Message----- From: Bryce Harrington [mailto:bryce@...1798...] Sent: vrijdag 22 februari 2008 3:07 To: Maximilian Albert Cc: Engelen, J.B.C. (Johan); inkscape-devel@lists.sourceforge.net; theadib@...1439... Subject: Re: [Inkscape-devel] crash on pdf/ps export
On Fri, Feb 22, 2008 at 01:36:19AM +0100, Maximilian Albert wrote:
J.B.C.Engelen@...1578... schrieb:
Heh, I just fixed this aswell last evening, but didn't
have internet.
I had the exact same fix :-)
Sometimes I wonder whether NULL pointer tests really solve these issues in all cases or if they sometimes just cover other
underlying
bugs (maybe the very fact that a NULL pointer is passed is wrong?). But in this case the fix seems to work fine; at least my exported pdf/ps files look good. :)
However, the crash unfortunately still occurs when one
presses "Close"
instead of "OK" in the pdf/ps options dialog which appears
after the
"Save file" dialog.
It's fairly traditional to use a NULL pointer to mean "not defined", so NULL pointer checks are quite standard and commonly used for controlling when certian logic should be invoked.
True, but in this case, I'm wondering whether it is better to do the check somewhere else. The function that tests for this NULL does nothing if it is NULL. Since it is a callback function (correct?) that is a reaction on pressing the OK button, maybe this shouldn't have been set as such in the first place.
I'll try to have a look into the other crash.
Johan
J.B.C.Engelen@...1578... schrieb:
... It's fairly traditional to use a NULL pointer to mean "not defined", so NULL pointer checks are quite standard and commonly used for controlling when certian logic should be invoked.
True, but in this case, I'm wondering whether it is better to do the check somewhere else. The function that tests for this NULL does nothing if it is NULL. Since it is a callback function (correct?) that is a reaction on pressing the OK button, maybe this shouldn't have been set as such in the first place.
I'll try to have a look into the other crash.
Johan
I totally agree that this NULL pointer check only prevents from crashing like try and catch block. I think that Teds commit still lacks in something (maybe some initialisation of variables?). But I am not familiar with this GTK functions...
Adib.
I fixed the crash, but I am not sure whether this introduced a bug somewhere else. (like a dialog not disappearing when it should)
Ted, could you please have a look at this?
Bryce, Josh: could one of you commit this to 0.46 ?
Thanks, Johan
-----Original Message----- From: Adib taraben [mailto:theadib@...1439...] Sent: vrijdag 22 februari 2008 9:51 To: inkscape-devel@lists.sourceforge.net; Ted Gould Cc: Engelen, J.B.C. (Johan); bryce@...1798...; Anhalter42@...173... Subject: Re: [Inkscape-devel] crash on pdf/ps export
J.B.C.Engelen@...1578... schrieb:
... It's fairly traditional to use a NULL pointer to mean "not
defined",
so NULL pointer checks are quite standard and commonly used for controlling when certian logic should be invoked.
True, but in this case, I'm wondering whether it is better
to do the check somewhere else. The function that tests for this NULL does nothing if it is NULL. Since it is a callback function (correct?) that is a reaction on pressing the OK button, maybe this shouldn't have been set as such in the first place.
I'll try to have a look into the other crash.
Johan
I totally agree that this NULL pointer check only prevents from crashing like try and catch block. I think that Teds commit still lacks in something (maybe some initialisation of variables?). But I am not familiar with this GTK functions...
Adib.
On Sat, 2008-02-23 at 21:35 +0100, J.B.C.Engelen@...1578... wrote:
I fixed the crash, but I am not sure whether this introduced a bug somewhere else. (like a dialog not disappearing when it should)
Yes, you did :) The problem is the life cycles of two objects need to be broken apart. The case that you're missing is when "Live Effects" get turned on and off. The parents change in those cases. I'm planning on fixing this tomorrow, making clear separation for the life cycles of these widgets so the bugs don't hide as easily.
--Ted
On Sat, Feb 23, 2008 at 01:11:51PM -0800, Ted Gould wrote:
On Sat, 2008-02-23 at 21:35 +0100, J.B.C.Engelen@...1578... wrote:
I fixed the crash, but I am not sure whether this introduced a bug somewhere else. (like a dialog not disappearing when it should)
Yes, you did :) The problem is the life cycles of two objects need to be broken apart. The case that you're missing is when "Live Effects" get turned on and off. The parents change in those cases. I'm planning on fixing this tomorrow, making clear separation for the life cycles of these widgets so the bugs don't hide as easily.
We'll hold off on integrating this until these remaining issues are sorted out.
Bryce
participants (7)
-
unknown＠example.com -
Adib taraben -
Bryce Harrington -
Maximilian Albert -
Rock Star -
Ted Gould -
the Adib