Closed Bug 493726 Opened 17 years ago Closed 17 years ago

Cairo/Qt backend fails to build -- again

Categories

(Core Graveyard :: Widget: Qt, defect)

Product:
Core Graveyard
Component:
Widget: Qt
Platform:
x86
Linux
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: tobias.hunger, Unassigned)

References

Details

Attachments

(2 files, 4 obsolete files)

Remove extra hunk Jeff did not like.
2.99 KB, patch
Details | Diff | Splinter Review
Clean up one unnecessary include.
480 bytes, patch
Details | Diff | Splinter Review
Attached patch Fix cairo qt build again (obsolete) — — Splinter Review
Hello! Cairo/Qt fails to build again after recent cairo updates. The attached patch fixes the issues.
Attachment #378314 - Flags: review?(mark.finkle)
Comment on attachment 378314 [details] [diff] [review] Fix cairo qt build again Just a drive by review >diff -r 068bfbbdabea configure.in >- if test ! -f "$LIBXUL_SDK/sdk/include/xpcom-config.h"; then >- AC_MSG_ERROR([$LIBXUL_SDK/sdk/include/xpcom-config.h doesn't exist]) >+ if test ! -f "$LIBXUL_SDK/include/xpcom-config.h"; then >+ AC_MSG_ERROR([$LIBXUL_SDK/include/xpcom-config.h doesn't exist]) I think a fix for this already landed > fi > > MOZ_ENABLE_LIBXUL=1 >@@ -7728,7 +7739,9 @@ > dnl ======================================================== > > QCMS_LIBS='$(DEPTH)/gfx/qcms/$(LIB_PREFIX)mozqcms.$(LIB_SUFFIX)' >+QCMS_CFLAGS='-I$(LIBXUL_DIST)/include' > AC_SUBST(QCMS_LIBS) >+AC_SUBST(QCMS_CFLAGS) > > dnl ======================================================== > dnl disable xul >diff -r 068bfbbdabea gfx/cairo/cairo/src/cairo-qpainter-surface.cpp >--- a/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp Tue May 19 17:44:23 2009 +1200 >+++ b/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp Tue May 19 14:11:06 2009 +0200 >@@ -41,6 +41,7 @@ > #include "cairo-types-private.h" > > #include "cairo-qpainter.h" >+ > #include <memory> > > #include <QtGui/QPainter> >@@ -740,7 +741,7 @@ > > cairo_traps_t traps; > cairo_int_status_t status; >- cairo_region_t region; >+ cairo_region_t * region(0); > > _cairo_traps_init (&traps); > status = (cairo_int_status_t) >@@ -781,12 +782,12 @@ > > QRegion qr; > >- n_boxes = _cairo_region_num_boxes(&region); >+ n_boxes = cairo_region_num_rectangles(region); > for (int i = 0; i < n_boxes; ++i) > { >- cairo_box_int_t box; >- _cairo_region_get_box(&region, i, &box); >- QRect r(box.p1.x, box.p1.y, box.p2.x - box.p1.x, box.p2.y - box.p2.y); >+ cairo_rectangle_int_t box; >+ cairo_region_get_rectangle(region, i, &box); >+ QRect r(box.x, box.y, box.width, box.height); > > if (0 == i) > clip_bounds = r; >@@ -796,7 +797,7 @@ > qr = qr.unite(r); > } > #endif >- _cairo_region_fini (&region); >+ _cairo_region_fini (region); > > qs->p->setClipRegion (qr, Qt::IntersectClip); > } else { >@@ -1406,7 +1407,7 @@ > qs->clip_bounds.height() > }; > >- _cairo_region_init_rect (&region, &rect); >+ _cairo_region_init_rectangle(&region, &rect); > > _cairo_surface_set_clip_region (qs->xlib_equiv, &region, ++qs->xlib_clip_serial); > >diff -r 068bfbbdabea gfx/thebes/src/Makefile.in >--- a/gfx/thebes/src/Makefile.in Tue May 19 17:44:23 2009 +1200 >+++ b/gfx/thebes/src/Makefile.in Tue May 19 14:11:06 2009 +0200 >@@ -173,8 +173,8 @@ > > include $(topsrcdir)/config/rules.mk > >-CXXFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) >-CFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) >+CXXFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) $(QCMS_CFLAGS) >+CFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) $(QCMS_CFLAGS) > > ifeq ($(MOZ_WIDGET_TOOLKIT),windows) > ifdef WINCE >diff -r 068bfbbdabea gfx/thebes/src/gfxPlatform.cpp > > #include "cairo.h" >-#include "qcms.h" This will affect non Qt backends, right? Will this break anyone? >diff -r 068bfbbdabea widget/src/qt/Makefile.in > LOCAL_INCLUDES += \ > -I$(topsrcdir)/widget/src/xpwidgets \ > -I$(srcdir) \ >+ -I$(topsrcdir)/gfx/qcms \ > $(NULL) indenting looks off
Comment on attachment 378314 [details] [diff] [review] Fix cairo qt build again Ugh, sorry for not snipping the patch chunks. r- so you can remove the configure.in changes.
Attachment #378314 - Flags: review?(mark.finkle) → review-
(In reply to comment #1) > (From update of attachment 378314 [details] [diff] [review]) > Just a drive by review > > >diff -r 068bfbbdabea configure.in > >- if test ! -f "$LIBXUL_SDK/sdk/include/xpcom-config.h"; then > >- AC_MSG_ERROR([$LIBXUL_SDK/sdk/include/xpcom-config.h doesn't exist]) > >+ if test ! -f "$LIBXUL_SDK/include/xpcom-config.h"; then > >+ AC_MSG_ERROR([$LIBXUL_SDK/include/xpcom-config.h doesn't exist]) > > I think a fix for this already landed Removed. > > #include "cairo.h" > >-#include "qcms.h" > > This will affect non Qt backends, right? Will this break anyone? Yes, but the gfxPlatform header already includes this file. I removed this include so that I do not needlessly have to fix up the include in two files. > >diff -r 068bfbbdabea widget/src/qt/Makefile.in > > > LOCAL_INCLUDES += \ > > -I$(topsrcdir)/widget/src/xpwidgets \ > > -I$(srcdir) \ > >+ -I$(topsrcdir)/gfx/qcms \ > > $(NULL) > > indenting looks off ... due to tabbing. Fixed.
Attachment #378314 - Attachment is obsolete: true
Attachment #379086 - Flags: review?
Attachment #379086 - Flags: review? → review?(mark.finkle)
Attachment #379086 - Flags: review?(mark.finkle) → review?(vladimir)
Comment on attachment 379086 [details] [diff] [review] Fix cairo qt build again (updated) Passing over to Vlad for the real review
Ping? It's been way over a week now without anything happening.
A small modification to Tobias' patch. Some things have changed related to config/autoconf.mk.in, so I think this patch being added here is closer to what we need.
Attachment #380122 - Flags: approval1.9.1?
Comment on attachment 380122 [details] [diff] [review] A revised patch for the Cairo Qt build This fixes a build failure of one of the gecko 1.9.1 backends, so it should be applied.
Flags: blocking1.9.1?
Ping? Could this patch please get commited? Thanks!
Not without a review on it, no. Ted, can you please take a look? Is Qt a supported backend?
It's not even listed on our supported build configurations page: https://developer.mozilla.org/En/Supported_build_configurations Without an active maintainer listed to help with issues, it's tough to keep it working, certainly. We sure don't need to block the release on this, we could take it in a dot release. (Most of it is NPOTB for our default builds.)
Flags: blocking1.9.1? → blocking1.9.1-
That page does not list GTK or any of the other toolkits either, so why should it list Qt? Anyway: The effected code is to be released. It does contain a crasher bug. At least four people agree on it (Me, James, Anton) and are effected by this bug. Mark (one of your mozilla devs!) had reviewed the bug. The fix is self contained (it is not seen by the compiler whenever a toolkit different from Qt is used at all!). So what is keeping you from applying it?
PS: We are running on linux/x86 (gcc) which is listed as a tier-1 platform in https://developer.mozilla.org/En/Supported_build_configurations
OK, Mark gave a review-, but his issues were fixed in the later updates to the patch.
Attachment #379086 - Attachment is obsolete: true
Attachment #379086 - Flags: review?(vladimir)
Attachment #380122 - Flags: review?(vladimir)
But it's not the default configuration for Linux/x86, and it's not built on our primary Tinderboxes, so it's not Tier 1. We could get this landed on trunk after review, sure. However, we're trying to ship Firefox 3.5 off of the 1.9.1 branch, and as such we're minimizing the amount of patches that land there at the moment. I'm sorry if this timing sucks for you, but after we ship 3.5 it should be easier to get this on branch. That being said, you don't have a patch here that's been granted review+ yet, and that's the first step to getting anything landed. I'm sorry if you feel that your patch has been languishing, but sometimes reviewers are busy and don't get to patches.
Attachment #380122 - Flags: review?(vladimir) → review?(jmuizelaar)
Comment on attachment 380122 [details] [diff] [review] A revised patch for the Cairo Qt build >diff --git a/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp b/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp >--- a/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp >+++ b/gfx/cairo/cairo/src/cairo-qpainter-surface.cpp >@@ -36,16 +36,17 @@ > > /* Get INT16_MIN etc. as per C99 */ > #define __STDC_LIMIT_MACROS > > #include "cairoint.h" > #include "cairo-types-private.h" > > #include "cairo-qpainter.h" >+ > #include <memory> > > #include <QtGui/QPainter> > #include <QtGui/QPaintEngine> > #include <QtGui/QPaintDevice> > #include <QtGui/QImage> > #include <QtGui/QPixmap> > #include <QtGui/QBrush> >@@ -735,17 +736,17 @@ _cairo_qpainter_surface_intersect_clip_p > qs->p->setClipRect (r, Qt::IntersectClip); > } else { > // Then if it's not an integer-aligned rectangle, check > // if we can extract a region (a set of rectangles) out. > // We use cairo to convert the path to traps. > > cairo_traps_t traps; > cairo_int_status_t status; >- cairo_region_t region; >+ cairo_region_t * region(0); I'd prefer this be initialized with '= NULL' instead of '(0)' as that fits with the cairo conventions better even though this is c++. > > _cairo_traps_init (&traps); > status = (cairo_int_status_t) > _cairo_path_fixed_fill_to_traps (path, fill_rule, tolerance, &traps); > if (status) { > _cairo_traps_fini (&traps); > return status; > } >@@ -776,32 +777,32 @@ _cairo_qpainter_surface_intersect_clip_p > qr = qr.unite(r); > } > _cairo_region_boxes_fini (&region, boxes); > #else > int n_boxes; > > QRegion qr; > >- n_boxes = _cairo_region_num_boxes(&region); >+ n_boxes = cairo_region_num_rectangles(region); > for (int i = 0; i < n_boxes; ++i) > { >- cairo_box_int_t box; >- _cairo_region_get_box(&region, i, &box); >- QRect r(box.p1.x, box.p1.y, box.p2.x - box.p1.x, box.p2.y - box.p2.y); >+ cairo_rectangle_int_t box; >+ cairo_region_get_rectangle(region, i, &box); >+ QRect r(box.x, box.y, box.width, box.height); > > if (0 == i) > clip_bounds = r; > else > clip_bounds = clip_bounds.united(r); > > qr = qr.unite(r); > } > #endif >- _cairo_region_fini (&region); >+ _cairo_region_fini (region); > > qs->p->setClipRegion (qr, Qt::IntersectClip); > } else { > // We weren't able to extract a region from the traps. > // Just hand the path down to QPainter. > QPainterPath qpath; > > if (_cairo_quartz_cairo_path_to_qpainterpath (path, &qpath, fill_rule) != CAIRO_STATUS_SUCCESS) >@@ -1401,17 +1402,17 @@ _cairo_qpainter_surface_show_glyphs (voi > cairo_region_t region; > cairo_rectangle_int_t rect = { > qs->clip_bounds.x() - qs->redir_offset.x(), > qs->clip_bounds.y() - qs->redir_offset.y(), > qs->clip_bounds.width(), > qs->clip_bounds.height() > }; > >- _cairo_region_init_rect (&region, &rect); >+ _cairo_region_init_rectangle(&region, &rect); > > _cairo_surface_set_clip_region (qs->xlib_equiv, &region, ++qs->xlib_clip_serial); > > _cairo_region_fini (&region); > } > > qs->xlib_has_clipping = qs->has_clipping; > qs->xlib_clip_bounds = qs->clip_bounds; These changes seem ok. >+CXXFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) $(QCMS_CFLAGS) >+CFLAGS += $(MOZ_CAIRO_CFLAGS) $(TK_CFLAGS) $(QCMS_CFLAGS) What are the qcms changes for? They seem like they should be a separate patch.
Attachment #380122 - Flags: review?(jmuizelaar) → review-
It seems that there is no need to play with any of the qcms things, so they have been removed. The use of zero as an initialiser has been replaced with NULL.
Attachment #380122 - Attachment is obsolete: true
Attachment #381514 - Flags: review?(jmuizelaar)
Attachment #380122 - Flags: approval1.9.1?
Ping? Any progress with the review/application of this patch?
Ping?
Look, sometimes people are busy. You pinged on the 5th, which was a friday, and in the early morning of the 8th (in Jeff and my timezone), a monday. You can't expect lightning-fast turnaround on patch reviews, ok? Jeff reviewed your previous patch, and I'm sure he'll get to your revised version soon. Please be patient.
I just want to make sure that this patch is not forgotten. It is pretty critical to me: I can not even build xulrunner without some heavy patching first! And this has been going on for weeks. I am under the impression that not even the upcoming release 1.9.1 will work! So I am getting desperate:-) Funny thing is that a crash I reported in the Qt code was fixed (even for Gecko 1.9.1!), while this patch that makes the code actually build in the first place is still sitting here in the bugtracker.
Comment on attachment 381514 [details] [diff] [review] A replacement patch for the Cairo Qt build > #include "cairo.h" >-#include "qcms.h" > There's still a qcms hunk here, with no reason for it given. r+ if this hunk is removed. Note: I wasn't cc'ed on the bug so I wasn't getting any of the pings regarding reviews.
Attachment #381514 - Flags: review?(jmuizelaar) → review+
There's absolutely no need for the inclusion of qcms.h in /gfx/thebes/src/gfxPlatform.cpp since /gfx/thebes/src/gfxPlatform.h already has it. This patch simply cleans that up.
Remove the one hunk Jeff did not like in the patch, leave the patch unchanged otherwise.
Attachment #381514 - Attachment is obsolete: true
This is the one hunk Jeff did not like in the build fix patch. It is a unnecessary include that gets removed, so it should get applied IMHO, but of course not as part of the build fix patch. So I am putting this as a separate patch here... don't want to open a new bugreport about it;-)
The mozilla-central tree is still closed at the moment, but we should be able to get this landed shortly after it opens. You can probably find someone to land it along with some other patches they have queued up.
Keywords: checkin-needed
Ted: Thanks for the feedback. #485579 was about the same piece of code as this one and did get applied. I really do not see why a crash fix can get applied and a build fix (which is required for the crash fix to even get build!) not. I think the magic phrase why this is applicable for commit even now is "NPOTB".
Yes, I would normally agree with you, but at the moment the tree is closed for release engineering maintenance before being reopened for checkins.
Blocks: 497858
So when will this patch finally land? The Qt backend is broken for weeks now!
No longer blocks: 497858
Landed now. I changed the patch to use '= NULL' instead of '(NULL)' as that's what I originally meant and missed it in the second review. http://hg.mozilla.org/mozilla-central/rev/baaf18eb76da Sorry it was broken for so long.
Status: NEW → RESOLVED
Closed: 17 years ago
Keywords: checkin-needed
Resolution: --- → FIXED
with the fixed code, build error on Ubuntu8.04 with Qt4.5.0 , mostly likely to be a Qt path configuration problem, but get confused because --with-qtdir is set clearly in .mozconfig ================================= //.mozconfig file: mk_add_options MOZ_CO_PROJECT=browser,xulrunner mk_add_options AUTOCONF=autoconf2.13 mk_add_options MOZ_OBJDIR=/home/myname/fun/gecko/qtxul mk_add_options MOZ_MAKE_FLAGS=-j4 ac_add_options --enable-application=browser ac_add_options --enable-default-toolkit=cairo-qt ac_add_options --with-qtdir="/usr/local/Trolltech/Qt-4.5.0" ac_add_options --enable-debug="-g3" ac_add_options --disable-optimize ac_add_options --enable-tests ac_add_options --disable-installer ac_add_options --disable-crashreporter ac_add_options --disable-javaxpcom ac_add_options --disable-printing #ac_add_options --disable-embedding-tests ac_add_options --disable-elf-dynstr-gc ================================= //error part of making: c++ -o moc_mozqwidget.o -c -I../../../dist/include/system_wrappers -include /home/myname/fun/gecko/mozilla-central/config/gcc_hidden.h -DMOZILLA_INTERNAL_API -DOSTYPE=\"Linux2.6\" -DOSARCH=Linux -D_IMPL_NS_WIDGET -D_BSD_SOURCE -I/home/myname/fun/gecko/mozilla-central/widget/src/xpwidgets -I/home/myname/fun/gecko/mozilla-central/widget/src/qt -I/home/myname/fun/gecko/mozilla-central/widget/src/qt -I. -I../../../dist/include -I../../../dist/include/nsprpub -I/home/myname/fun/gecko/qtxul/dist/include/nspr -I/home/myname/fun/gecko/qtxul/dist/include/nss -fPIC -fno-rtti -fno-exceptions -Wall -Wpointer-arith -Woverloaded-virtual -Wsynth -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -Wcast-align -Wno-invalid-offsetof -Wno-long-long -pedantic -fno-strict-aliasing -fshort-wchar -pthread -pipe -DDEBUG -D_DEBUG -DDEBUG_myname -DTRACING -g3 -DQT_SHARED -I/usr/include/qt4 -I/usr/include/qt4/QtGui -I/usr/include/qt4/QtNetwork -I/usr/include/qt4/QtUiTools -I/usr/include/qt4/QtCore -I/usr/include/glib-1.2 -I/usr/lib/glib/include -DMOZILLA_CLIENT -include ../../../mozilla-config.h -Wp,-MD,.deps/moc_mozqwidget.pp moc_mozqwidget.cpp moc_mozqwidget.cpp:14:2: error: #error "This file was generated using the moc from 4.5.0. It" moc_mozqwidget.cpp:15:2: error: #error "cannot be used with the include files from this version of Qt." moc_mozqwidget.cpp:16:2: error: #error "(The moc has changed too much.)" moc_nsAppShell.cpp c++ -o moc_nsAppShell.o -c -I../../../dist/include/system_wrappers -include /home/myname/fun/gecko/mozilla-central/config/gcc_hidden.h -DMOZILLA_INTERNAL_API -DOSTYPE=\"Linux2.6\" -DOSARCH=Linux -D_IMPL_NS_WIDGET -D_BSD_SOURCE -I/home/myname/fun/gecko/mozilla-central/widget/src/xpwidgets -I/home/myname/fun/gecko/mozilla-central/widget/src/qt -I/home/myname/fun/gecko/mozilla-central/widget/src/qt -I. -I../../../dist/include -I../../../dist/include/nsprpub -I/home/myname/fun/gecko/qtxul/dist/include/nspr -I/home/myname/fun/gecko/qtxul/dist/include/nss -fPIC -fno-rtti -fno-exceptions -Wall -Wpointer-arith -Woverloaded-virtual -Wsynth -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -Wcast-align -Wno-invalid-offsetof -Wno-long-long -pedantic -fno-strict-aliasing -fshort-wchar -pthread -pipe -DDEBUG -D_DEBUG -DDEBUG_myname -DTRACING -g3 -DQT_SHARED -I/usr/include/qt4 -I/usr/include/qt4/QtGui -I/usr/include/qt4/QtNetwork -I/usr/include/qt4/QtUiTools -I/usr/include/qt4/QtCore -I/usr/include/glib-1.2 -I/usr/lib/glib/include -DMOZILLA_CLIENT -include ../../../mozilla-config.h -Wp,-MD,.deps/moc_nsAppShell.pp moc_nsAppShell.cpp moc_nsAppShell.cpp:14:2: error: #error "This file was generated using the moc from 4.5.0. It" moc_nsAppShell.cpp:15:2: error: #error "cannot be used with the include files from this version of Qt." moc_nsAppShell.cpp:16:2: error: #error "(The moc has changed too much.)" make[6]: *** [moc_mozqwidget.o] Error 1 make[6]: *** Waiting for unfinished jobs.... /home/myname/fun/gecko/mozilla-central/widget/src/qt/nsFilePicker.cpp: In member function \u2018virtual nsresult nsFilePicker::Show(PRInt16*)\u2019: /home/myname/fun/gecko/mozilla-central/widget/src/qt/nsFilePicker.cpp:244: warning: format \u2018%s\u2019 expects type \u2018char*\u2019, but argument 2 has type \u2018QChar*\u2019 make[6]: *** [moc_nsAppShell.o] Error 1 make[6]: Leaving directory `/home/myname/fun/gecko/qtxul/widget/src/qt' make[5]: *** [libs] Error 2 make[5]: Leaving directory `/home/myname/fun/gecko/qtxul/widget/src' make[4]: *** [libs] Error 2 make[4]: Leaving directory `/home/myname/fun/gecko/qtxul/widget' make[3]: *** [libs_tier_gecko] Error 2 make[3]: Leaving directory `/home/myname/fun/gecko/qtxul' make[2]: *** [tier_gecko] Error 2 make[2]: Leaving directory `/home/myname/fun/gecko/qtxul' make[1]: *** [default] Error 2 make[1]: Leaving directory `/home/myname/fun/gecko/qtxul' make: *** [build] Error 2
This is an Ubuntu problem, not a problem of this patch or anything to do with Mozilla's build system. I have seen on Ubuntu 8.04 that the output of pkg-config is incorrect, even if you specify PKG_CONFIG_PATH to pick up .pc files from elsewhere.
Following up on this: (a bit off topic, discussing Mozembed-QT build, seg fault) QT build of mozilla-central pass on Ubuntu 8.04 with QT4.5.0 Mozembed-QT build pass, too. BUT ./testqmozembed has a seg fault. Any suggestions/hints will be of great appreciation. ================================= //GDB outputs showing the seg fault: Breakpoint 3, InitEmbedding (aProfilePath=0x0, aComps=0x0, aNumComps=0) at ../common/EmbeddingSetup.cpp:273 273 if (sProfileDir && !sProfileLock) { (gdb) l 268 if (!dirExists) { 269 sProfileDir->Create(nsIFile::DIRECTORY_TYPE, 0700); 270 } 271 272 // Lock profile directory 273 if (sProfileDir && !sProfileLock) { 274 rv = XRE_LockProfileDirectory(sProfileDir, &sProfileLock); 275 NS_ENSURE_SUCCESS(rv, rv); 276 } 277 (gdb) bt #0 InitEmbedding (aProfilePath=0x0, aComps=0x0, aNumComps=0) at ../ common/EmbeddingSetup.cpp:273 #1 0x08050fce in MozView (this=0x80c00d8) at ../common/embed.cpp:264 #2 0x0804ec22 in Private (this=0x80c00d8, aQMozView=0x80c3b50) at QMozView.cpp:68 #3 0x0804e696 in QMozView (this=0x80c3b50, parent=0xbfce2f70, flags=0) at QMozView.cpp:128 #4 0x0804cb49 in MyQMozView (this=0x80c3b50, parent=0xbfce2f70, flags=0) at test.cpp:50 #5 0x0804cc99 in MyBrowser (this=0xbfce2f70, parent=0x0, flags=0) at test.cpp:70 #6 0x0804d026 in main (argc=1, argv=0xbfce3044) at test.cpp:131 (gdb) n 274 rv = XRE_LockProfileDirectory(sProfileDir, &sProfileLock); (gdb) s nsCOMPtr<nsILocalFile>::operator nsILocalFile* (this=0x8069e04) at /home/myname/fun/gecko/qtxul/dist/bin//../include/nsCOMPtr.h: 790 790 return get(); (gdb) bt #0 nsCOMPtr<nsILocalFile>::operator nsILocalFile* (this=0x8069e04) at /home/myname/fun/gecko/qtxul/dist/bin//../include/nsCOMPtr.h: 790 #1 0x080546a8 in InitEmbedding (aProfilePath=0x0, aComps=0x0, aNumComps=0) at ../common/EmbeddingSetup.cpp:274 #2 0x08050fce in MozView (this=0x80c00d8) at ../common/embed.cpp:264 #3 0x0804ec22 in Private (this=0x80c00d8, aQMozView=0x80c3b50) at QMozView.cpp:68 #4 0x0804e696 in QMozView (this=0x80c3b50, parent=0xbfce2f70, flags=0) at QMozView.cpp:128 #5 0x0804cb49 in MyQMozView (this=0x80c3b50, parent=0xbfce2f70, flags=0) at test.cpp:50 #6 0x0804cc99 in MyBrowser (this=0xbfce2f70, parent=0x0, flags=0) at test.cpp:70 #7 0x0804d026 in main (argc=1, argv=0xbfce3044) at test.cpp:131 (gdb) n Program received signal SIGSEGV, Segmentation fault. 0x090ea4a6 in ?? () ======================================
Just in case FYI, Mozembed-QT is here: https://wiki.mozilla.org/Embedding/NewApi/QtLinux
hui xue: I think it would make sense to report this as a separate bug. The xulrunner code obviously builds for you, so it is not related to this bug. It further is closed, so people will tend to ignore your problem while it is reported here only. Best Regards, Tobias
Can this bug be fixed also in branch mozilla-1.9.1 ? cairo-qpainter-surface.cpp: In function 'cairo_int_status_t _cairo_qpainter_surface_show_glyphs(void*, cairo_operator_t, cairo_pattern_t*, cairo_glyph_t*, int, cairo_scaled_font_t*, int*)': cairo-qpainter-surface.cpp:1391: error: invalid conversion from 'int' to 'cairo_text_cluster_flags_t' cairo-qpainter-surface.cpp: In function 'void _cairo_image_surface_write_to_ppm(cairo_image_surface_t*, const char*)': cairo-qpainter-surface.cpp:1798: warning: deprecated conversion from string constant to 'char*' cairo-qpainter-surface.cpp:1800: warning: deprecated conversion from string constant to 'char*' cairo-qpainter-surface.cpp: At global scope: cairo-qpainter-surface.cpp:129: warning: 'g_dump_path' defined but not used cairo-qpainter-surface.cpp:130: warning: 'timer_start_val' defined but not used cairo-qpainter-surface.cpp:155: warning: 'const char* _opstr(cairo_operator_t)' defined but not used
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: