[Sebastian, Will, Alexander] Problems compiling kdebase-4.3 due to trig

[Volker Krause]

On Tuesday 24 November 2009 19:56:12 Alexander Neundorf wrote:
> On Tuesday 24 November 2009, Volker Krause wrote:
> > On Monday 23 November 2009 22:43:55 Alexander Neundorf wrote:
> > > On Sunday 22 November 2009, Alexander Neundorf wrote:
> > > > On Friday 20 November 2009, Sebastian Trueg wrote:
> > > > > are you sure there isn't something wrong with your setup?
> > > > > After all, the header is installed by kdelibs/nepomuk/core/ui
> > > >
> > > > Let's see tomorrow when kdelibs builds again.
> > >
> > > So, kdebase is not building since more than a week now.
> > > The problem is this in kdelibs/CMakeLists.txt:
> > >
> > > if(Soprano_FOUND AND SOPRANO_PLUGIN_RAPTORPARSER_FOUND
> > >    AND SOPRANO_PLUGIN_REDLANDBACKEND_FOUND)
> > >   set(HAVE_NEPOMUK true)
> > >   add_subdirectory(nepomuk)
> > > endif(Soprano_FOUND AND SOPRANO_PLUGIN_RAPTORPARSER_FOUND
> > >       AND SOPRANO_PLUGIN_REDLANDBACKEND_FOUND)
> > >
> > > This part is skipped on my box, which is ok, since this is optional,
> > > otherwise cmake should fail and report that it can't build kdelibs
> > > because RAPTORPARSER and REDLANDBACKEND have not been found.
> > >
> > > But as it is, it succeeds, and skips nepomuk, and by this does not
> > > install e.g. ./nepomuk/core/ui/kmetadatatagwidget.h .
> > > But this is used later on in kdepim, which then fails:
> > > http://my.cdash.org/viewBuildError.php?buildid=38741
> > >
> > > So, what do we do ?
> > >
> > > Is SOPRANO_PLUGIN_RAPTORPARSER_FOUND and
> > > SOPRANO_PLUGIN_REDLANDBACKEND_FOUND really optional or should we make
> > > it required ?
> > >
> > > Or should the code be built only optionally ?
> > > Then it needs proper checks.
> > >
> > > As it is now, it is broken.
> >
> > as far as KDE PIM is concerned (which the above mentioned build error
> > referes to), Nepomuk is a hard dependency for all practical purposes,
> > this is just not yet reflected in the buildsystem correctly everywhere
> > (it is mostly a runtime dependency anyway).
>
> What does that mean ? Would nepomuk still work if the generated files (from
> raptor/trig) were put into svn and generated only on demand by those who
> know what they are doing ?
> Is the redland and raptor/trig stuff also necessary at runtime or only at
> buildtime ?

ah, sorry, I forgot about the code generators. KDE PIM uses both the 
ontology2class and rcgen tool. I don't know the details, but I think the 
redland/raptor stuff is a runtime dependency of these tools (and thus a build 
time dependency for us). So, yes, putting the generated code into svn could 
solve some of these problems. But given the amount of code as well as the 
number of places where it is used I would leave that as a last resort, eg. if 
some of those tools are not available on one of our supported platforms.

I have no idea about the exact runtime dependencies of Nepomuk itself, I can 
only speculate that eg. the Nepomuk ontology loader might need some of this 
as well, given the ontology based code generators need it.

regards
Volker
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20091125/f915deb0/attachment.sig>