May 2006 - Inkscape Devel - Inkscape Mailman

Need help on bug repro for #1497837
by Jon A. Cruz 31 May '06

31 May '06

Can anyone out there reproduce bug 1497837? [1497837] crash with GTK+ 2.8.17 http://sourceforge.net/tracker/? func=detail&aid=1497837&group_id=93438&atid=604306 I can see where the final crash point might be caused by a few different things, but running with --g-fatal-warnings could catch the first step in things breaking down and give another helpful backtrace. Thanks.

1 0

static rpm depends on glitz
by Nicu Buculei 30 May '06

30 May '06

The static rpm for 0.44pre0 refuses to install on my system because of a missing dependency: glitz. Is it really necessary or just a build mistake?

3 3

gcc 4.1.1 compile warnings
by Jon A. Cruz 30 May '06

30 May '06

I was looking over things a little more, and I *think* I know a better approach to solve the warnings. Once I started looking at things, I realized that in at least some of the cases we have an SPObject* we need to end up with. Conceptually that's a GTK+/Glib object, not a C++ one, so we just need to work in the right paradigm My first thought was something like this for sp-flowtext.cpp: void* rawptr = 0; SPObject* source_obj = 0: ... group->layout.getSourceOfCharacter( it, &rawptr, &span_text_start_iter); source_obj = reinterpret_cast<SPObject*>(rawptr); However, Ralf pointed out that using a reinterpret_cast<> itself triggered that warning. After I had that kicking around in my head for a while, I finally realized what I was missing: SP_OBJECT() We need to force an unknown pointer to be one to an SPObject. However, that's a Glib object, not a base C++ one. And we already have that macro to coerce an arbitrary pointer into a pointer to an SPObject. Plus it does all its magic in C-land, not C++ land. So using it should clear up the warnings, and also is the "proper" GTK+ thing to do. However... since I'm only on gcc 3.3 here, can someone else try this to see if it cleans things up on gcc 4.1.x? Thanks (I'm expecting that at the worst the SP_OBJECT macro will have the warning pop back up. But if it does then we'll have a single point in the codebase to change to fix everything, instead of needing to maintain every place that other function is used) Index: src/sp-flowtext.cpp =================================================================== --- src/sp-flowtext.cpp (revision 12065) +++ src/sp-flowtext.cpp (working copy) @@ -558,21 +558,23 @@ if (set_y) sp_repr_set_svg_double(span_tspan, "y", anchor_point [NR::Y]); - union { SPObject *op; void *vp; } source_obj; + SPObject* source_obj = 0; + void* rawptr = 0; Glib::ustring::iterator span_text_start_iter; - group->layout.getSourceOfCharacter(it, &source_obj.vp, &span_text_start_iter); - gchar *style_text = sp_style_write_difference ((SP_IS_STRING(source_obj.vp) ? source_obj.op->parent : source_obj.op)->style, group->style); + group->layout.getSourceOfCharacter(it, &rawptr, &span_text_start_iter); + source_obj = SP_OBJECT(rawptr); + gchar *style_text = sp_style_write_difference ((SP_IS_STRING(source_obj) ? source_obj->parent : source_obj)->style, group->style); if (style_text && *style_text) { span_tspan->setAttribute("style", style_text); g_free(style_text); } - if (SP_IS_STRING(source_obj.vp)) { - Glib::ustring *string = &SP_STRING(source_obj.vp)- >string; + if (SP_IS_STRING(source_obj)) { + Glib::ustring *string = &SP_STRING(source_obj)->string; union { SPObject *op; void *vp; } span_end_obj; Glib::ustring::iterator span_text_end_iter; group->layout.getSourceOfCharacter(it_span_end, &span_end_obj.vp, &span_text_end_iter); - if (span_end_obj.op != source_obj.op) { + if (span_end_obj.op != source_obj) { if (it_span_end == group->layout.end()) { span_text_end_iter = span_text_start_iter; for (int i = group- >layout.iteratorToCharIndex(it_span_end) - group- >layout.iteratorToCharIndex(it) ; i ; --i)

2 3

0.44pre1
by Aaron Spike 30 May '06

30 May '06

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I've just uploaded the tarball for 0.44pre1. Please package and test. Thanks again to rwst for help with distcheck. Aaron Spike -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (MingW32) iD8DBQFEfHzz1wsJX4VXgboRAimrAJ43wL1+gZ55lHdiR/aIMGFK5r4v2wCgwQo/ wwVEaqvE56XetZ76L4u7iho= =3mOq -----END PGP SIGNATURE-----

1 0

translation file problem
by Ralf Stephan 30 May '06

30 May '06

Can the translation managers please make sure that such errors do not get in the way of packagers? I only found this after 30 min of make distcheck. This should be found earlier. Please resolve this. ralf ../../po/lt.po:2674: `msgid' and `msgstr' entries do not both end with '\n' ../../po/lt.po:2692: `msgid' and `msgstr' entries do not both end with '\n' ../../po/lt.po:2708: `msgid' and `msgstr' entries do not both end with '\n' /usr/local/bin/msgfmt: found 3 fatal errors make[3]: *** [lt.gmo] Error 1 make[2]: *** [all-recursive] Error 1 make[1]: *** [all] Error 2 make: *** [distcheck] Error 2

1 1

pre1 distcheck
by Aaron Spike 30 May '06

30 May '06

Getting an error with `make distcheck` currently. I put it up at the same spot again. http://www.ekips.org/comp/inkscape/inx/distcheck/distcheck_stderr.txt Aaron Spike

2 1

Inkscape 0.44 Bug Status [#05]
by Bryce Harrington 30 May '06

30 May '06

Fixed Bugs Since 5/24/06 Owner ID Title Pts ---------------------------------------------------------------------- matipha 1496681 Build issue for zh_CN.po 9 rwst 1495283 Crashes instead of executing scripts 9 JFL 1488446 Zero Install and --enable-binreloc 6 adarovs 1495414 Whiteboard module login restrictions 6 acspike 1249746 effects must add stuff to current layer, not root 3 bbyak 1496421 "Break apart" too slow on complex paths 3 joncruz 1495856 cannot hide/lock layer without making it current 3 kees 1489407 Update the Inkscape logo in the website header 3 mjwybro 1448811 Connector: assertion failure when move empty group 9 bbyak 1496114 whitespace in rules breaks lindenmeyer effect 3 bbyak 496081 cosmetic issues in effect parameters dialog 3 cyreve 1495265 (WIN) Unicode display fails in latest build 6 ulferik 1474575 1.234e-5 notation used in LaTeX-files 6 kees 1495310 no parameters for output extensions 6 acspike 1244836 use newer glib2 on official build system 9 joncruz 1280117 layers are gone switching from linux to windows 6 joncruz 1235798 Changing layer order doesn't update layer selector 3 joncruz 1493995 Resize handle on bottom-right of window missing 6 joncruz 1495333 quick layer selector doesn't lock in layer dialog 6 bryce 1490928 Error message not translated 6 bbyak 933482 dashes/ markers do not survive stroke-to-path 6 mjwybro 1495284 OSX: compilation error: missing CREATE dir paths 6 bryce 1490042 [WEBSITE] RSS-Feed sent as text/plain 3 cyreve 1493680 no compile with FreeType-2.2.1-2 9 ---------------------------------------------------------------------- Total (goal: 200) 135 Must-Fix List for 0.44.0 ------------------------ *rwst 1467073 (DONE) make distcheck failing *cyreve 1493680 (DONE) no compile with FreeType-2.2.1-2 yipdw 1489175 Win2K crash: entrypoint FreeAddrInfo not found ishmal 1489961 Autopackage: crash on save as ODG *acspike 1244836 use newer glib2 on official build system Bryce On Wed, May 24, 2006 at 08:36:14PM -0700, Bryce Harrington wrote: > Hi all, > > Now that we are in Feature Freeze, attention turns to bug fixing. > > As we've done before, we'll establish a point goal - 200pts. Each fixed > bug between today and release will count towards that score. Low > priority bugs score 3pts, medium 6pts, and high 9pts. > > When we've hit that list, *AND* closed all the Must-Fix bugs for 0.44.0, > we will move into the Hard Freeze phase, and give control over to the > release wardens to finalize the release. > > > Needs Fixed in 0.44.x Stable Series > ----------------------------------- > cyreve 1494337 Fonts/Win: this TTF font crashes Inkscape (pango error) > rwst 1467137 Shrinking a pattern causes lock-up > cth03 1488128 Snap: Aligning (Snapping) to Grid or Guides does not > work > joncruz 1418122 Conf corruption when opening files with non-UTF encoded > joncruz 1370772 WinXP: Preferences.XML can't be accessed > mjwyb 1448811 Connector: assertion failure when moving empty group > ishmal 1395505 0.43 crash at startup on Win98se > mental 1441003 Misc: flowRoot error with inkscape's svg > cyreve 1448618 Fonts: weight variants selection > rwst 1493580 SuSE: CVS builds but will not run on Suse 10.0 > cth03 1429049 Snap: near-freeze in ObjectSnapper::_findCandidates > with ~1000 obj > > Needs Fixed for 0.45 > -------------------- > dwyip 1464771 whiteboard: cannot connect to jabber server then > crashses > cth03 1431560 Snap: Crash at resizing/snapping to guidelines > > Needs Owner in Order to be Prioritized > -------------------------------------- > ---- 1484076 potrace freeze with multiple color scan > ---- 1475189 Extension ignored by inkscape > ---- 1468393 square/circle/polygon/spiral and linked offset > misbehaviours > ---- 1491646 WIN - Win32 extensions does not call delivered Python > ---- 1491925 gradients referenced from masked object disappear > ---- 1490430 [INFO NEEDED] Crash in libcairo.dll > ---- 1489794 crash with composite > ---- 1459502 Crash on imported image update (during Gimp save) > ---- 1349761 Inkboard: inkscape crashes when trying to connect > > > Note that I've deliberately kept the list of must-fix bugs for 0.44.0 > very lean, so that we can release sooner rather than later. If you > disagree and have a bug that you think really must go there, then please > speak up now, and explain why it should be a release blocker. > > Some of the rationale I used for de-ranking bugs: Difficulty confirming > the bug. Bug appears to affect just the reporter. Bug doesn't cause > crash or data loss. Crash is due to something very unusual (like a > specific font, or a directory with unicode characters, or selecting > >1000 nodes while snapping is on, etc.) Bug is in a feature not enabled > by default (i.e. inkboard). > > Also note that this list doesn't accept bugs unless they have an owner. > This is strictly for pragmatic reasons - without someone volunteering to > take care of it, we may end up waiting a very long time to close it. > In past releases, we've often ended up simply having to release with > these sorts of bugs anyway, so I figure, why accept them as Must Fixes > in the first place? ;-) > > Bryce > > > ------------------------------------------------------- > All the advantages of Linux Managed Hosting--Without the Cost and Risk! > Fully trained technicians. The highest number of Red Hat certifications in > the hosting industry. Fanatical Support. Click to learn more > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 > _______________________________________________ > Inkscape-devel mailing list > Inkscape-devel(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/inkscape-devel ------------------------------------------------------- All the advantages of Linux Managed Hosting--Without the Cost and Risk! Fully trained technicians. The highest number of Red Hat certifications in the hosting industry. Fanatical Support. Click to learn more http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 _______________________________________________ Inkscape-devel mailing list Inkscape-devel(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-devel ----- End forwarded message ----- ----- End forwarded message ----- ----- End forwarded message ----- ----- End forwarded message -----

1 0

Turn Off Inkboard
by Aaron Spike 30 May '06

30 May '06

Ok. Let's turn of inkboard in all builds meant for distribution for the remainder of the release process. I've turned it off for the autopackages and the windows builds. Is there anywhere else that it was on by default? Mac? Aaron Spike

2 1

Difference between SPObject::requestModified and requestDisplayUpdate?
by Bryce Harrington 30 May '06

30 May '06

Hey Mental, Could you explain what the difference is between SPObject::requestModified and SPObject::requestDisplayUpdate? I notice that sp_stop_set() calls the former, whereas sp_marker_set() uses the latter, but looking at the code it's not clear why they are calling separate things... (I'm trying to sort out a fix for bug #1298718) Bryce

1 0

Re: [Inkscape-user] Can't get autpackage of .44
by Bryce Harrington 29 May '06

29 May '06

On Mon, May 29, 2006 at 02:44:31PM +0200, Josef Vybiral wrote: > On Mon, 29 May 2006 14:29:18 +0200, Abrolag > <abrolag@...58...> wrote: > > >When I try to download the latest autopackage build, instead of starting > >up the download manager it tries to load into the browser window. This > >happens with both FireFox 1.5 and Opera 8.54. > > > >What am I doing wrong? > > > > Hi, try to right click the link to the .package file and use "Save link > target as" option in the context menu that appears. The problem is that > server returns the file as simple text and browser reads it in this way. Hmm, thanks for pointing this out, this looks like a problem. Fortunately, it's easy to solve... we just need to specify a mime-type for the .package files. Could someone with login access to modevia look into this? I think it's simply a matter of adding an .htaccess file with a line like: AddType application/x-autopackage .package But that's probably not the right mime-type... Mike Hearn, is there a registered mime-type for autopackage files? (If not, there should be!) I notice that the packages on the autopackage.org site sometimes downloads as type "BIN", which makes me think there isn't an associated type yet? Bryce

2 3