as a followup to remote-specific meta-data, now i'm thinking of injecting data in the
git-annex branch. i'm still a little hesitant because optimising lookupkey mentions that i need to "delete or update
.git/annex/index" when doing so. yet I also saw in the source code there is a
.git/annex/index.lock file that is being used to avoid concurrent access to that file - is that something that should be watched out for? how do I manage that lockfile? is it mandatory to stage files into the index at first?
maybe i'm just not familiar enough with how git operates internally myself. i've reviewed this section of the progit book again: http://git-scm.com/book/en/v2/Git-Internals-Git-Objects - and i feel i have again some handle on how to do those things, but it seems like a major hurdle to go through just to change a single setting in one git-annex metadata file... wouldn't
enableremote be sufficient here? I figured to append custom fields to a remote, i could do:
git annex enableremote myremoteuuid "$(git cat-file -p git-annex:remote.log | grep myremoteuuid) extratag=true"
Would that work at all? Should i operate directly on the git-annex branch myself instead?