Why not handle this with thin meta-repos? Take this example:
multimedia.git is a meta-repo, therefore it need not contain any code. What if the only thing it contained was a list of submodule paths and a script to clone them all if they have not already been cloned? Since this only really ever has to be once, it could be a viable solution. The script need not be very complicated either. ( for subrepo in `cat subrepos.list`; do if [[ ! -d $subrepo ]] ; then git clone $BASE_URL/$subrepo.git ; fi; done ) git handles errors nicely enough that we wouldn't have to bother handling them ourselves, and it would be really simple to maintain the submodules, and really simple to move submodules around (edit the text files) This is what I actually think we should do, since as long as the scripts are REALLY short and simple, transparency with git will be kept (so people understand what is happening) and yet the directory structure can be set up quickly. Also a partial download is recoverable (since any successful clone will be skipped next run, but a failed one would be re-run) Please respond with comments here, I want to know what the shortcomings of this kind of system would be. I also believe that this will help enable multiple workflows, the two main ones being the "all-at-once" and the "pick- and-choose".
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Kde-scm-interest mailing list [email protected] https://mail.kde.org/mailman/listinfo/kde-scm-interest
