Repos really are entities, like nodes - which means having a system name for them (and all the requirements system names have) is really not so sensible. Much better to have a nickname for them, analogous to a node's title. There's no reason we can't continue to load by name with such a change - just you'd no longer be guaranteed getting a single repo back if loading by name.
This'll make things a little easier for project/sandboxes on d.o, since right now we need to artificially namespace them all.
| Comment | File | Size | Author |
|---|---|---|---|
| #8 | 0001-task-1016646-by-marvil07-sdboyer-Make-versioncontrol.patch | 2.14 KB | marvil07 |
| #4 | non-special-repo-names-v1.patch | 2.84 KB | marvil07 |
| #3 | non-special-repo-names-v0.patch | 2.12 KB | marvil07 |
Comments
Comment #1
sdboyer commentedGonna push this to sprint 9. It's all under-the-hood, anyway.
Comment #2
sdboyer commentedThis'd be a good one for marvil07 to take, actually, if he's got time. Leaving in sprint 8 for now, will leave it up to him to delay if he wants.
Comment #3
marvil07 commentedAgree about making it non-unique, but not really sure if we want to have name optional
Here a patch that do both things.
Comment #4
marvil07 commentedtest fix
Comment #6
marvil07 commentedtest bot can not handle contrib testing with module dependencies :-/
Comment #7
sdboyer commentedLooks good here. Are we sure there aren't any cases where we use the name as a unique lookup? I can't think of any - the only one I remember is making the drush integration take names as arguments for the repo sync, but that's fine as it's designed to be able to handle multiple repos at once.
Comment #8
marvil07 commentedAfter analyzing the output of
git grep -C4 "name"I think there are not ;-)Here the patch I committed without the hunk for making it optional, as discussed with sdboyer.