Hello Brian, On Wednesday 08 February 2006 18:34, Brian J. Murrell wrote: > Hello, > > I work for Cluster File Systems. One of my current projects is to > (finally) eliminate our use of a custom version of quilt (the b_cfs > branch). It seems that "stock quilt" comes very close to allowing us to > do this with the exception of the "setup" step. > > As per a previous conversation with Andreas, please find the patch > below. It allows one to do this: > > $ quilt setup -l path/to/series/file -d path/to/patches/directory > > This setup mode assumes that the series file is a list of patch files > and that the patch files listed in there are in the > path/to/patches/directory. These setup switches create symbolic links > in . to the specified series file and patches directory. i.e.: > > lrwxrwxrwx 1 brian brian 53 Feb 8 12:04 series -> > ../b1_4/lustre/kernel_patches/series/2.6-rhel4.series lrwxrwxrwx 1 brian > brian 39 Feb 8 12:04 patches -> > ../b1_4/lustre/kernel_patches/patches/// > > One can then do quilt push/pop operations normally.
I still don't get it: the setup command currently sets up a source tree from a set of tarballs and a set of patches by untaring the tarballs and creating symlinks back to the patches directory. If a spec file is passed, it additionally figures out which the tarballs and patches are. If you are working against a cvs tree, how does the source tree get created that you want to have the series and patches links created in? Don't you use a script for creating that tree from the cvs? If it's a script, then why can't that script create the symlinks in the first place, instead of adding some weird options to the setup command? Thanks, Andreas _______________________________________________ Quilt-dev mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/quilt-dev
