libreoffice FTBFS in saucy

Bug #1182170 reported by Benjamin Drung
10
This bug affects 1 person
Affects Status Importance Assigned to Milestone
libreoffice (Ubuntu)
Fix Released
High
Björn Michaelsen

Bug Description

libreoffice 1:4.0.2-0ubuntu1 (and the prepared 1:4.0.3-0ubuntu1 update) fails to build from source in saucy. pbuilder build log attached.

Revision history for this message
Benjamin Drung (bdrung) wrote :
Benjamin Drung (bdrung)
Changed in libreoffice (Ubuntu):
importance: Undecided → High
Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

@xnox: Are you tweaking on the toolchain update right now? Is your current workstate public somewhere?

Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

Taking over from xnox, found his state of work at http://people.canonical.com/~xnox/repo/

Changed in libreoffice (Ubuntu):
assignee: nobody → Björn Michaelsen (bjoern-michaelsen)
Revision history for this message
Dimitri John Ledkov (xnox) wrote :

Yeah, my bits are at:
dget http://people.canonical.com/~xnox/repo/libreoffice_4.0.2-0ubuntu2.dsc

Please note the hsqldb transition / new upstream release is in saucy-proposed now.

Changed in libreoffice (Ubuntu):
status: New → In Progress
Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

I build that and can reproduce the test failure, which is a crasher to make things more interesting. Triaged it down to the sw.XMLImporter test in sw/qa/unoapi/sw.sce, which is the testcode at:

 https://gerrit.libreoffice.org/gitweb?p=core.git;a=blob;f=qadevOOo/tests/java/mod/_sw/XMLImporter.java;h=c6707c59bf11c00723d65a3a05b1f88d5f86d98c;hb=HEAD

Unfortunately this is not a direct issue, as the crash happens when the external java process "undocks", that is when the bridge is torn down, so the issue long like some fun to debug lifecycle issue ...

Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

so to further triage this one down to this interface test:

 http://opengrok.libreoffice.org/xref/core/qadevOOo/tests/java/ifc/xml/sax/_XDocumentHandler.java

and of that, the startDocument test is enough to trigger the problem on teardown.
In SwXMLImport::startDocument it seems the first line triggering the trouble is:

 setTextInsertMode( xInsertTextRange );

at http://opengrok.libreoffice.org/xref/core/sw/source/filter/xml/xmlimp.cxx#642

Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

Test failure is:
REPRODUCIBLE even with --without-system-boost
NOT REPRODUCABLE on libreoffice master@28f2af669f01cd304994f1907de0b61245b3d37a (CommitDate: Sun May 26 15:14:07 2013 +0100)

However, I unfortunately find no obvious commit responsible for the fix.

 git rev-list --count libreoffice-4.0.2.2..master

shows 10329 commits in this range, making this a 16 step bisect search, so Im off for some CPU burnage ...

Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

eh, 14 step not 16 step. Not that it makes much of a difference.

Revision history for this message
Björn Michaelsen (bjoern-michaelsen) wrote :

Discussed this with seb128:
- as upstream 4.1 doesnt exhibit this breaker
- LibreOffice beta1 is out and beta2 will follow soon, instead of bisecting the fix for backporting, just go directly for uploading beta2 in saucy and simply disable this test in the 4.0 package.

Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package libreoffice - 1:4.0.2-0ubuntu5

---------------
libreoffice (1:4.0.2-0ubuntu5) saucy; urgency=low

  * Add libcommons-logging-java BD, to work around brokenness when doing an
    arch build.
 -- Iain Lane <email address hidden> Fri, 31 May 2013 23:24:52 +0100

Changed in libreoffice (Ubuntu):
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.