Goswin von Brederlow wrote: > Andreas Beckmann <deb...@abeckmann.de> writes: > >> Package: ia32-libs-tools >> Version: 20 >> Severity: normal ... >> 1. /usr/lib/ia32-libs-tools/fetch calls 'apt-get update' but fails. ... >> Is it really neccessary to call apt-get update from fetch (and every >> time? Wouldn't it be OK to fail during fetching packages if the lists >> are not up-to-date as it would happen with apt-get install? >> The user should just remember to run update first or live with the >> consequences otherwise. > > I think that is acceptable and actually preferable.
How will this work for build-all? A single update takes much longer than the corresponding package build. >> 2. fetching the binary/source package fails if there is more than one >> version available, e.g. because sources.list contains stable, ... > All this is the fetch script and I see the problem. It tries to > downloads the source for every version of the bzip2 binary it > knows. And you do not have a deb-src entry for every deb entry, so not > every version does have sources. And even with sources for all > versions the create will later complain about multiple *.dsc files. All my test happend in pbuilder environments and you are right, there are no deb-src lines at all :-) > Afaik for apt-get source neither -t testing not /testing works. But if > I use the above idea to fetch binaries first and scan them then that > can be supported by fetch and create. -t does work, /testing does not (which is a little bit confusing and already reported: #414105). >> There is a script missing that builds converted packages for the command >> line parameter(s) together with a changes file so that all this can be >> uploaded to a local repository. > > How about create <pkg>, build <pkg>, build-all? Sounds good. Regards, Andreas -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
