[PATCH]][Fwd: Changes in Strigi caused build problems in 4.1 BRANCH]

Andreas Pakulat apaku at gmx.de
Fri Aug 29 14:59:11 BST 2008


On 29.08.08 05:49:20, James Richard Tyrer wrote:
> Alex Merry wrote:
> <SNIP>
> > This is due to a change in Strigi.  The Strigi currently in kdesupport will 
> > work with KDE trunk but not KDE 4.1.  I think Strigi 0.5.x should work with 
> > 4.1, but not trunk.
> > 
> After considerable research I again report that it doesn't work -- it
> will not compile.
> 
> The problem should be stated clearly.
> 
> KDELibs 4.1 BRANCH will not build against Strigi 0.5 BRANCH.
> 
> The reason is that revision 852727 changed line 43 in:
> 
> branches/KDE/4.1/kdelibs/kio/kio/dummyanalyzers/dummyanalyzers.cpp
> 
> from:
> 
> signed char analyze(Strigi::AnalysisResult&, InputStream*) {
> 
> to:	
> 
> char analyze(Strigi::AnalysisResult&, InputStream*) {
> 
> AND revision 849001 changed line 37 in:
> 
> branches/strigi/0.5/src/streamanalyzer/streamendanalyzer.h
> 
> from:
> 
> virtual char analyze(Strigi::AnalysisResult& idx, InputStream* in)=0;
> 
> to:
> 
> virtual signed char analyze(Strigi::AnalysisResult& idx, InputStream* in)=0;
> 
> NOTE: the strigi-0.5.10 does not contain this change.

Then the error is in strigi, it shouldn't do BiC changes in a stable
release branch. If you look at svn log its easily visible that Jos
simply took the wrong revision of trunk/kdesupport/strigi for the
branch. So this change should be reverted in the strigi 0.5 branch

Jos if you're reading this can you please revert revision 849001 in the
0.5 branch. That should fix the build problems on platforms where char
!= signed char.

> This type of a change will always cause problems.  The only way to avoid 
> them is to have KDE-4.x.y check to see that it has the exact version of 
> Strigi that it needs.

Yeap, kdelibs 4.1.x needs a check for strigi < 0.6, dunno if anybody is
already writing that.

Andreas

-- 
Today is the first day of the rest of your life.
 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<




More information about the kde-core-devel mailing list