Issue 65207

Summary: loading the OOo specification template using http:// crashes the office
Product: App Dev Reporter: jogi
Component: scriptingAssignee: jogi
Status: CLOSED NOT_AN_OOO_ISSUE QA Contact: Unknown <non-migrated>
Severity: Trivial    
Priority: P2 CC: christian.jansen, issues, joerg.skottke, tino.rachui
Version: 3.3.0 or older (OOo)Keywords: regression
Target Milestone: ---   
Hardware: All   
OS: All   
URL: http://specs.openoffice.org/collaterals/template/OpenOffice-org-Specification-Template.ott
Issue Type: DEFECT Latest Confirmation in: ---
Developer Difficulty: ---
Issue Depends on:    
Issue Blocks: 65112    
Attachments:
Description Flags
Problem DEV has solved none

Description jogi 2006-05-09 14:35:44 UTC
Loading
http://specs.openoffice.org/collaterals/template/OpenOffice-org-Specification-Template.ott
or also through file system crashes the office if you ACTIVATE the macros
(medium security settings (default)). If you deny activating macros the document
will be loaded.
Comment 1 jogi 2006-05-09 14:36:52 UTC
adding tr, cj to issue
Comment 2 jogi 2006-05-09 14:39:35 UTC
Evaluated on Solaris SPARC with a SRC680m156 (patched from m124) where it was okay.
Comment 3 joerg.skottke 2006-05-10 06:59:28 UTC
cc me
Comment 4 ab 2006-05-10 07:39:05 UTC
Started
Comment 5 joerg.skottke 2006-05-10 07:51:45 UTC
Created automated Testcase: f_basic_spectemplate.bas::tBasicSpecTemplate to be
run on every build. The test loads the file from the webpage (according to cj
the URL is static) and accepts executing macros.
Comment 6 ab 2006-05-10 15:04:53 UTC
ab->mba: As discussed, SfxObjectShell::PostActivateEvent_Impl()
posts events too early due to changed VCL timer handling.
Comment 7 Mathias_Bauer 2006-05-11 13:30:54 UTC
As discovered the bug is triggered by unexpected timer events that happen
because nowadays VCL checks for expired timers while painting and then
immediately calls the handler procudure. 
As discussed we will make sure that timer events still will be processed in the
"regular" Yield() calls only by posting a User Event when an expired timer is
detected while painting a window.
Comment 8 philipp.lohmann 2006-05-11 13:32:54 UTC
fixed in CWS mbapp3fix1
Comment 9 Mathias_Bauer 2006-05-15 13:36:01 UTC
The fix provided by PL is a Windows only fix - it fixes the problem that macros
bound to an "OnLoad" or "OnCreate" event don't work on Windows if a window
(floating toolbar etc.) is shown before the macro gets executed. The reason for
this bug is so urgent (uncontrollable time event processing can lead to crashes
in code that is not reentrant safe) that we should treat it as a showstopper for
2.0.3.

The crash on Solaris couldn't be reproduced on the milestones m167-m169.

Comment 10 Mathias_Bauer 2006-05-15 16:39:02 UTC
Now the macro that is bound to the "create document" event should be executed on
Windows even if a floating toolbar is open.

re-open issue and reassign to jsi@openoffice.org
Comment 11 Mathias_Bauer 2006-05-15 16:39:09 UTC
reassign to jsi@openoffice.org
Comment 12 Mathias_Bauer 2006-05-15 16:39:14 UTC
reset resolution to FIXED
Comment 13 jogi 2006-05-16 09:19:01 UTC
Created attachment 36501 [details]
Problem DEV has solved
Comment 14 jogi 2006-05-16 09:20:09 UTC
Attached the issue, which happens only on Win32, development has fixed but that
was not my problem I had with it. AS is investigating on the stack.
Comment 15 andreas.schluens 2006-05-16 11:33:57 UTC
AS->ABI: Please ignore existing comments inside this issue regarding timer handling in 
vcl added by PL. It describes another issue !

re-open issue and reassign to abi@openoffice.org
Comment 16 andreas.schluens 2006-05-16 11:34:04 UTC
reassign to abi@openoffice.org
Comment 17 andreas.schluens 2006-05-16 11:34:11 UTC
reset resolution to FIXED
Comment 18 andreas.schluens 2006-05-16 11:35:16 UTC
.
Comment 19 andreas.bille 2006-05-16 13:09:06 UTC
started
Comment 20 andreas.bille 2006-05-16 14:29:28 UTC
ABI->JS:
For me not reproducable on a m169, neither in the respin nor in a m124 with
patch from m169.
Comment 21 jogi 2006-05-18 13:30:29 UTC
Tested on OOCm1 (patched) and also okay for me.
Comment 22 jogi 2006-05-18 13:31:05 UTC
closing.