[pkg-bioc] pbuilder does not find newly built packages

[David Vernazobres]

On Sun, Apr 29, 2007 at 06:59:18PM +0200, Steffen Moeller wrote :
> On Sunday 29 April 2007 18:45:20 David Vernazobres wrote:
> > On Sun, Apr 29, 2007 at 12:53:36AM +0200, Steffen Moeller wrote :
> > > Hi, it is time for me to say good night. I found the last issue with my
> > > pbuilder-hook that works on bioc while I am compiling cran :-/  We do not
> > > lose too much if we make the hook run on all our build directories
> > > regardless what repository we are working with, do we? Well, maybe we do
> > > for 1700 packages ...
> >
> > Well, we need them during the pbuilder build. We need to regenerate all
> > the Packages.gz at each step, if not we missed the previously build deb
> > and we can't go further for the build package process with pbuilder.
> >
> > > Here my changed hook (also added the ()s)
> > >
> > > #!/bin/sh
> > > for m in cran bioc-1.9 omegahat; do
> > >   (cd /nfshome/moeller/debian_moeller/debian/alioth/bioc/${m}/ &&
> > >   /usr/bin/dpkg-scanpackages builds /dev/null | gzip > builds/Packages.gz
> > > ) done
> > > /usr/bin/apt-get update
> >
> > Hum, sorry, but what is the difference with the
> > ${PBuilder_BuildPlaceHook}/D70apt script?
> It is your script, only that all the directories are in and not only the one 
> that was specified in the r_pkg_prepare.pl  (bioc for me).
humm, right. It was not a bug, but I wrote it like that, because I am
mainly building bioc|cran in a parallel way (in 3-4 loop). As we do not
managed the dependency in a global way. Most of the time with pbuilder,
is opening the base.tgz, installing stuff,... The limitation is the
disk access, not the processor.


> > And I may be tired, but 
> > what's the point for the ()s ?
> Without the () the cwd changes and the next call to cd in the for loop will be 
> unvalid since we are in the subdir now. ... which does not matter while only 
> working on CRAN :-)
Damm, this one is a bug ! I will commit the patch right now !

Cheers,
david