The buildworld has also exposed a major flaw in the r-base* packaging. The
builds logs for…
http://www.snaggledworks.com/fink/buildworld/2014-07-24/logs/cran-slam-r30.log
and
http://www.snaggledworks.com/fink/buildworld/2014-07-24/logs/cran-rjava-r31.log
show the missing dependencies on gcc49-compilers and libicu48-dev (among
others from
/sw/Library/Frameworks/R.framework/Resources/etc/Makeconf's linkage of
LIBS = -lpcre -llzma -lbz2 -lz -lm -liconv -licuuc -licui18n
The debian developers solve this in their r-base-dev package with
dependencies on the required *-dev packages…
https://packages.debian.org/sid/r-base-dev
Unfortunately the use of BuildDependsOnly: True in the r-base*-dev
split-off prohibits this in fink. So it would appear that the only solution
is to explicitly add those missing *-dev dependencies to the BuildDepends
in every single cran package in libs/rmods. Ugh.
Jack
ps It makes one wonder if fink couldn't be modified to special case the
BuildDependsOnly: True to disable it for the *-dev split-off itself. That
is that Depends on *-dev which use BuildDependsOnly: True are ignored for
the case of their presence in a *-dev which has already has
BuildDependsOnly: True itself since that will recursively force the
BuildDependsOnly: True on the other -dev packages to be honored despite
their presence in the Depends of r-rbase*-dev.
On Wed, Aug 20, 2014 at 7:55 AM, Hanspeter Niederstrasser <
hanspe...@snaggledworks.com> wrote:
> >
> http://www.snaggledworks.com/fink/buildworld/2014-07-24/out/filters.xml
> >
> > but this gives me a 404 error.
>
> That's fixed now.
>
> > In any case, I fixed some of the issues you pointed out (in sdl2,
> fflas-ffpack, mc, the-silver-searcher) but now am mostly left with
> "failures/project/insufficient_permission" failures, but I am not quite
> sure what the failure indicates. Perhaps this:
> >
> > cp: /sw/share/info/dir.bak: Permission denied
> >
> > I think this is caused by the install-info tool and the fact that the
> affected packages (e.g. grep) use the "InfoDocs" fields. I don't think a
> package author can do much about that, can they? As such, perhaps you could
> check if the above line is the only one with "permission denied", and in
> that case, demote the "failure" to a "warning" ?
>
> Yeah, I've downgraded those to a warning. fink-bld takes care of any
> writing being done directly into %p.
>
> Hanspeter
>
>
>
> ------------------------------------------------------------------------------
> Slashdot TV.
> Video for Nerds. Stuff that matters.
> http://tv.slashdot.org/
> _______________________________________________
> Fink-devel mailing list
> Fink-devel@lists.sourceforge.net
> List archive:
> http://news.gmane.org/gmane.os.apple.fink.devel
> Subscription management:
> https://lists.sourceforge.net/lists/listinfo/fink-devel
>
------------------------------------------------------------------------------
Slashdot TV.
Video for Nerds. Stuff that matters.
http://tv.slashdot.org/
_______________________________________________
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel
