Bug 81756 - Crasher in Format:Page... menu item
Summary: Crasher in Format:Page... menu item
Status: RESOLVED DUPLICATE of bug 81496
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Calc (show other bugs)
Version:
(earliest affected)
4.4.0.0.alpha0+ Master
Hardware: All macOS (All)
: medium critical
Assignee: Not Assigned
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: mab4.4
  Show dependency treegraph
 
Reported: 2014-07-25 18:24 UTC by uli
Modified: 2014-08-22 12:08 UTC (History)
4 users (show)

See Also:
Crash report or crash signature:


Attachments
Crashreporter log (13.92 KB, application/zip)
2014-07-25 18:24 UTC, uli
Details

Note You need to log in before you can comment on or make changes to this bug.
Description uli 2014-07-25 18:24:18 UTC
Created attachment 103461 [details]
Crashreporter log

With a spreadsheet open, picking Format:Page... produceds a spinning beach ball for a few seconds, then crash. This appears to affect most 4.0.0.alpha0 nightlies incl. the one I just pulled yesterday. It also affects the PPC built bu Doug Mencken/OSU OSL.
I will append a crash log (zipped).
Comment 1 Jorendc 2014-07-30 19:13:53 UTC
Not reproducible, tested using Mac OSX 10.9 with LibreOffice 4.3.0.4 rc. Also not repro on Ubuntu 14.04.
Comment 2 uli 2014-07-30 19:51:24 UTC
(In reply to comment #1)
> Not reproducible, tested using Mac OSX 10.9 with LibreOffice 4.3.0.4 rc.
> Also not repro on Ubuntu 14.04.

No surprise: The bug is triggered in Lo 4.4.0.0.alpha0 on OS X 10.6.8. That is, Lo four-point-four. It is NOT present in 4.3.0.1 (last 4.3 I have). Very reproducible with Libreoffice nightly:
TinderBox: MacOSX-x86@49-TDF, Branch:master, Time: 2014-07-11_18:39:38

I will check against a more recent lightly soon.
Comment 3 Jorendc 2014-07-30 19:57:29 UTC
(In reply to comment #2)
> (In reply to comment #1)
> No surprise: The bug is triggered in Lo 4.4.0.0.alpha0 on OS X 10.6.8. That
> is, Lo four-point-four. It is NOT present in 4.3.0.1 (last 4.3 I have).

Thanks for being friendly and not mentioning you can't reproduce this in 4.3. I only have access to 4.3.0.4 test setup right now, so I left the comment I can't in 4.3.0.4. Not more, not less. Accept it, if you don't mention it, someone else will. Not sure where you problem starts with that? Feel free to discuss this on the mailinglist or on IRC.
Comment 4 uli 2014-07-30 20:14:09 UTC
(In reply to comment #3)
> (In reply to comment #2)
> > (In reply to comment #1)
> > No surprise: The bug is triggered in Lo 4.4.0.0.alpha0 on OS X 10.6.8. That
> > is, Lo four-point-four. It is NOT present in 4.3.0.1 (last 4.3 I have).
> 
> Thanks for being friendly and not mentioning you can't reproduce this in
> 4.3. I only have access to 4.3.0.4 test setup right now, so I left the
> comment I can't in 4.3.0.4. Not more, not less. Accept it, if you don't
> mention it, someone else will. Not sure where you problem starts with that?
> Feel free to discuss this on the mailinglist or on IRC.

My apologies; in the OP I explicitly mention the version where I found the bug. But yes, I could have mentioned that it appears to be a new one introduced with the 4.4 version and does not appear to be present in 4.3 (or 4.2, for that matter). Also note that I found it in the Spreadsheet module only. It is not present in the Writer module. I do not know about the other modules.

My main aim is that it should be known to those working on 4.4, which otherwise seems to be quite a nice version (at least for me so far), and that it does not get closed before being addressed. It is a rather serious issue.

I don't have access to OS X newer than 10.6.8 so cannot comment on those systems. Bugzilla does not seem to allow to specify versions of OS X.

Peace?

Uli
Comment 5 Jorendc 2014-07-30 20:18:18 UTC
(In reply to comment #4)
> My apologies;

No worries ;-). More than accepted.

> It is a rather serious issue.
It is, therefore we have to try to narrow the range and make sure where it is introduced.

> I don't have access to OS X newer than 10.6.8 so cannot comment on those
> systems. Bugzilla does not seem to allow to specify versions of OS X.

True

> Peace?

Peace :).

Sure, will test this bug later this week :-).

Kind regards,
Joren
Comment 6 Jorendc 2014-07-30 21:18:12 UTC
Reproducible, tested using Mac OSX 10.9 with LibreOffice Version: 4.4.0.0.alpha0+ Build ID: 998ab5c405963accf47afc69a6bf04e045f6f71f
TinderBox: MacOSX-x86@49-TDF, Branch:master, Time: 2014-07-22_03:13:14

Marking as NEW
Comment 7 Douglas Mencken 2014-07-30 22:11:43 UTC
I confirm this regression.

It's probably caused by commit 7d9bb549d ( https://gerrit.libreoffice.org/gitweb?p=core.git;a=commitdiff;h=7d9bb549d498d6beed2c4050c402d09643febdfa ),
where we have svx/source/unodraw/unobrushitemhelper.cxx with that setSvxBrushItemAsFillAttributesToTargetSet method, which fails on one of rToSet.Put.

0   libsvllo.dylib                	0x010df254 SfxItemPool::Put(SfxPoolItem const&, unsigned short) + 612
1   libsvllo.dylib                	0x010ea64c SfxItemSet::Put(SfxPoolItem const&, unsigned short) + 476
2   libsvxcorelo.dylib            	0x2029d6cc setSvxBrushItemAsFillAttributesToTargetSet(SvxBrushItem const&, SfxItemSet&) + 2396
Comment 8 Douglas Mencken 2014-07-31 16:33:06 UTC
It affects not only x86. (Changed to "All".)

It happens every time.

Maybe set to "blocker"?
Comment 9 Douglas Mencken 2014-07-31 16:45:40 UTC
I did

--- a/svx/source/unodraw/unobrushitemhelper.cxx
+++ b/svx/source/unodraw/unobrushitemhelper.cxx
@@ -132,6 +132,7 @@ void setSvxBrushItemAsFillAttributesToTargetSet(const SvxBrushItem& rBrush, SfxI
             rToSet.Put(XFillTransparenceItem(nGraphicTransparency));
         }
     }
+#if 0
     else
     {
         // GPOS_NONE == rBrush.GetGraphicPos() && 0xff == rBrush.GetColor().GetTransparency(),
@@ -146,8 +147,11 @@ void setSvxBrushItemAsFillAttributesToTargetSet(const SvxBrushItem& rBrush, SfxI
         const Color aColor(rBrush.GetColor().GetRGBColor());
 
         rToSet.Put(XFillStyleItem(drawing::FillStyle_NONE));
+#if 0
         rToSet.Put(XFillColorItem(OUString(), aColor));
+#endif
     }
+#endif
 }
 
and got:

0   libsvxcorelo.dylib            	0x16638ac4 drawinglayer::primitive2d::createNewSdrFillAttribute(SfxItemSet const&) + 100
1   libsvxcorelo.dylib            	0x165e8d90 drawinglayer::attribute::SdrAllFillAttributesHelper::SdrAllFillAttributesHelper(SfxItemSet const&) + 96
2   libcuilo.dylib                	0x1886fb3c SvxPageDescPage::ResetBackground_Impl(SfxItemSet const&) + 2444
3   libcuilo.dylib                	0x188705dc SvxPageDescPage::Reset(SfxItemSet const*) + 1996
4   libsfxlo.dylib                	0x05283528 SfxTabDialog::ActivatePageHdl(TabControl*) + 1048
5   libsfxlo.dylib                	0x05283974 SfxTabDialog::Start_Impl() + 628
6   libsfxlo.dylib                	0x05283b0c SfxTabDialog::Execute() + 44
7   libscuilo.dylib               	0x14c7b1cc ScAbstractTabDialog_Impl::Execute() + 28
8   libsclo.dylib                 	0x154859ac ScDocShell::ExecutePageStyle(SfxViewShell&, SfxRequest&, short) + 316
9   libsclo.dylib                 	0x157a176c ScTabViewShell::Execute(SfxRequest&) + 1612
...

So it's cui/source/tabpages/page.cxx or underlining design.

The same cui dialog works for writer. But writer pages have both footer, header (or absense of them). They might be filled, even with transparent gradient, yeah, it's cool.
But is it fully applicable to Calc pages (given that the same @@ -1280,7 +1280,7 @@ void SvxPageDescPage::ResetBackground_Impl(const SfxItemSet& rSet) is used)?

I know this patch is adopted from Apache, but anyway...
Comment 10 Julien Nabet 2014-08-05 20:39:19 UTC
Miklos: following Douglas' comment (https://bugs.freedesktop.org/show_bug.cgi?id=81756#c7), I thought you might be interested in this one.
Comment 11 Eike Rathke 2014-08-22 12:08:20 UTC

*** This bug has been marked as a duplicate of bug 81496 ***