On 26/7/10 22:24, ~suv wrote:

1. Good news :)

fixed annoying crashes from 0.47 (which many users still experience and complain about, using the official stable release version 0.47):

2. annoying open bugs that crash inkscape (in 0.47 and trunk) - sorted

by bug number, not 'Importance' or 'Heat':

3) issues without one main report but recurrently mentioned in a number of reports

Note: I'm not a developer, the chance of having confused issues in the lists below is high. Any help sorting out duplicate or invalid reports in the bug tracker would be very appreciated.

a) Memory/GC related crashes during PNG export/PDF import/SVG open|save: unknown if fixable/fixed for all platforms; spread across various reports, many likely related to #168914 (which has seen most activity). I am not sure if bug #168914 better be reopened for Inkscape until verified to be addressed for all affected platforms (reports in duplicates and comments are not limited to win32)):

keywords: Too many heap sections GC Warning: Repeated allocation of very large block GC Warning: Out of Memory! Returning NIL! terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc

-- Too many heap sections 168914 Too many heap sections (win32, Debian, Ubuntu Lucid) https://bugs.launchpad.net/inkscape/+bug/168914 226565 Too many heap sections: Increase MAXHINCR or MAX_HEAP_SECTS (Gentoo) https://bugs.launchpad.net/inkscape/+bug/226565 262243 Too many heap sections, while Ungrouping (win32) https://bugs.launchpad.net/inkscape/+bug/262243 427003 Fatal error in gc crash on save (win32) https://bugs.launchpad.net/inkscape/+bug/427003 444940 Too many heap sections (win32) https://bugs.launchpad.net/inkscape/+bug/444940 455928 Inkscape (Windows) crashes under load (win32) https://bugs.launchpad.net/inkscape/+bug/455928 481252 Emergency save destroys config file (linux) https://bugs.launchpad.net/inkscape/+bug/481252 510380 Too Many Heap Sections in inkscape 0.47 (win32) https://bugs.launchpad.net/inkscape/+bug/510380

-- GC Warning / std::bad_alloc 168738 Out of memory on 5MB svg file opening (Ubuntu Feisty) https://bugs.launchpad.net/inkscape/+bug/168738 168937 [needfile] Out of memory in openstreetmap tiles@...623... project (linux) https://bugs.launchpad.net/inkscape/+bug/168937 214859 inkscape crashes, blank error message box, corrupted backup and original save (PCLinuxOS) https://bugs.launchpad.net/inkscape/+bug/214859 367779 Out of memory crash while importing large pdf (Ubuntu 7.10) https://bugs.launchpad.net/inkscape/+bug/367779 308589 Inkscape crashes when large file has been opened and zoomed some times (Debian, Ubuntu) https://bugs.launchpad.net/inkscape/+bug/308589 394228 inkscape crashes with bad alloc on conversion from svg to png (win32) https://bugs.launchpad.net/inkscape/+bug/394228 392421 Crash when open *.ai more tham 500mb (Ubuntu 9.04) https://bugs.launchpad.net/inkscape/+bug/392421

-- memory leaks 169102 Memory leaking. Perhaps use Boehm atomic malloc more. (linux) https://bugs.launchpad.net/inkscape/+bug/169102 597689 Memory leak when showing/hiding embedded images. (win32, osx) https://bugs.launchpad.net/inkscape/+bug/597689 608944 Enormous memory consumption (memory leak?) (win32, Ubuntu, osx) https://bugs.launchpad.net/inkscape/+bug/608944

-- GLib-ERROR **: gmem.c175: failed to allocate... 450898 Win64: gmem memory allocation fails at 1 MB (win32, osx) https://bugs.launchpad.net/inkscape/+bug/450898

b) crashes or failure on opening SVG file due to encoding errors (various reports not yet really triaged and sorted into main report + duplicates, or determined unrelated)

keywords: parser error : PCDATA invalid Char value parser error : Input is not proper UTF-8, indicate encoding ! <?xpacket begin=

184904 Pasting text containing control character creates unreadable SVG https://bugs.launchpad.net/inkscape/+bug/184904 291416 SVG from imported PDF cannot be reopened, characters error https://bugs.launchpad.net/inkscape/+bug/291416 297701 Cannot open SVG file that was saved from a PDF opened with "Embed Images" option set https://bugs.launchpad.net/inkscape/+bug/297701 369861 Unable to open previously imported pdf file https://bugs.launchpad.net/inkscape/+bug/369861 496831 slow and cannot save when trying to resize eps files https://bugs.launchpad.net/inkscape/+bug/496831 499257 saves svg file, that it can't read afterwards https://bugs.launchpad.net/inkscape/+bug/499257 513895 inkscape 0.47 does not open own svg https://bugs.launchpad.net/inkscape/+bug/513895 ...

c) failure to open PDF/PS files (mainly win32)

keywords: ?

585927 Inkscape crashes on importing of some PS https://bugs.launchpad.net/inkscape/+bug/585927 492389 crash when user opens a PDF https://bugs.launchpad.net/inkscape/+bug/492389 494164 Open PDF crashes before "import settings" window appears https://bugs.launchpad.net/inkscape/+bug/494164 498814 Error "Inkscape encountered an internal error and will close now" when opening eps from gnuplot. https://bugs.launchpad.net/inkscape/+bug/498814 412609 pdf import crashes in vista https://bugs.launchpad.net/inkscape/+bug/412609 ...

~suv

On 27/7/10 05:30, ~suv wrote:

One I definitely forgot to list: not a crash (but perceived as such by many users), critical (data loss):

On 26/7/10 22:24, ~suv wrote:

...

2. annoying open bugs that crash inkscape (in 0.47 and trunk) - sorted

by bug number, not 'Importance' or 'Heat':

285267 [Win32] Save As/Save a Copy not shown https://bugs.launchpad.net/inkscape/+bug/285267

Summary for bug #285267 https://bugs.launchpad.net/inkscape/+bug/285267 ------------------------------------------------------------------------- 2008-10-18 Windows XP Pro SP3 SVN 19900 2009-05-27 windowsxp sp3 developmental versions 2009-08-03 Windows XP SP3 0.47pre1 2009-12-11 Win XP Prof SP3 Inkscape 0.47 r22583 2009-12-07 Windows XP SP2/SP3 Inkscape 0.47 2009-12-29 WIN-XP pro 47-3 2010-01-10 Windows XP Professional SP3 Inkscape 0.47 r22583 2010-01-15 Windows XP Professional SP3 2010-03-02 0.47, R9126 2010-03-09 xp pro sp3 R9170 2010-06-06 Win 7 Home Prem. x64 [1] Inkscape 0.47 2010-06-19 Windows XP SP3 0.47, r9492 2010-07-03 Win XP Inkscape 0.47 2010-07-27 WinXP SP2 9596, 9640 2010-07-27 winXP, not with windows 7

similar problems also seen with Scribus 1.3.5 SVN OpenOffice (Writer)

Issue A identified as 'custom theme'

caused by: 'modded' version of XP with some 3rd party themes UXTheme Patcher reinstalling XP without the internet explorer core components Fresh Installation (No Internet Explorer Core Components) workarounds: change desktop theme to Windows classic Removing UXTheme resolved the bug use Win classic theme disable the windows themeing (aka uxtheme, aka Luna, etc) switch back to the Classic Windows Theme (odd size)

Issue B identified as 'incorrect window size'

caused by: inkscape somehow manages to create a window with zero width. workarounds: autohotkey script (force arbitrary size and position) -------------------------------------------------------------------------

Question: Is this a blocker for 0.48 (or fixed in a dot release)?

The bug seems to be limited to Windows XP [1].

Statistics about usage of Windows XP: According to Wikipedia [2], «As of the end of June 2010, Windows XP is the most widely used operating system in the world with a 54.6% market share, having peaked at 76.1% in January 2007.»

Support status of Windows XP: Extended Support until 8 April 2014 (only Service Pack 3 x86 and Service Pack 2 x64)

------------------------------------------------------------------------- [1] only one comment reports the bug is present in Windows 7 too: https://bugs.launchpad.net/inkscape/+bug/285267/comments/31

[2] Wikipedia: Windows XP http://en.wikipedia.org/wiki/Windows_XP

Defensive patch that might fix this. Also attached to the bug.

Regards, Krzysztof

W dniu 5 sierpnia 2010 09:28 użytkownik Jon Cruz <jon@...18...> napisał:

One note, though. I see that the patch is using a "MAX" macro. Now that Inkscape is using C++, we should try to use std::max() instead. Hmm... and the placement of the "+ 1" outside of the max seems a little odd.

I have a suspicion that either MoveWindow breaks when the width is the same as before, or GetWindowRect returns bogus values. The "+1" is to avoid the first case.

Oh, and now that I think of it, if it is helpful to cap the width, it might also be handy to watch the height. That's one area where GTK programs get dinged for not respecting the current dynamic 'usable' area, etc.

Height was not a problem in the bug report, but it might make some sense to add it. I think that native dialogs should be removed in 0.49 because they are a maintenance nightmare, and make implementing features like the "embed" checkbox in the import dialog more difficult than it should be.

Regards, Krzysztof

On Aug 5, 2010, at 12:14 PM, Krzysztof Kosiński wrote:

W dniu 5 sierpnia 2010 09:28 użytkownik Jon Cruz <jon@...18...> napisał:

...

One note, though. I see that the patch is using a "MAX" macro. Now that Inkscape is using C++, we should try to use std::max() instead. Hmm... and the placement of the "+ 1" outside of the max seems a little odd.

I have a suspicion that either MoveWindow breaks when the width is the same as before, or GetWindowRect returns bogus values. The "+1" is to avoid the first case.

Ahh... then it really needs to be broken out. That is, collapsing things to a single sub-expression inside of a function call hides what the intent is, and can contribute to bugs.

There are also problems with using MAX or std::max at all. This is really not the case where that is desired, as the value we're checking against should not be the large one we use for a default.

But probably the most important thing is to try to explicitly test your suspicions. At the lest, separating the work-arounds so that an end user could easily comment out just one of them would really help with testing.

...

Oh, and now that I think of it, if it is helpful to cap the width, it might also be handy to watch the height. That's one area where GTK programs get dinged for not respecting the current dynamic 'usable' area, etc.

Height was not a problem in the bug report, but it might make some sense to add it. I think that native dialogs should be removed in 0.49 because they are a maintenance nightmare, and make implementing features like the "embed" checkbox in the import dialog more difficult than it should be.

Yes, height was also an issue. Someone explicitly called out the zero-height problem in comment #32 there.