Allow the shed_ operations to consume git URLs instead of paths.#174
Allow the shed_ operations to consume git URLs instead of paths.#174jmchilton merged 6 commits intogalaxyproject:masterfrom
Conversation
create, upload, download, diff, and lint projects directly from Github - closes galaxyproject#169. Inputs starting with ``git+`` (e.g. git+/path/to/repo.git) with will have the ``git+`` stripped and passed directly to the ``git`` command-line. Alternatively URLs starting with `git:` will be passed as is to the ``git`` command-line. This seems to roughly correspond to ``pip``.
Allow the shed_ operations to consume git URLs instead of paths.
There was a problem hiding this comment.
Assuming we are only targeting github/bitbucket it would be faster to pull the latest tarball. With all the test data over time a git pull can become a bottle-neck.
There was a problem hiding this comment.
So this line is pulling Galaxy down for test/serve when used with --install_galaxy - which used to use a tar ball by default but now does a clone so that it can do incremental updates on each call very quickly.
It sounds like you mean for the actual shed stuff though - which could be optimized somewhat. I don't think we should just download the tars though because I would like to use the git metadata during upload (#170) and because this will need to support submodules at some point. I do think we could do clone with --depth=1 probably to speed things up though paired with creating a directory to cache repositories between calls (mirroring how Galaxy is quickly cloned).
There was a problem hiding this comment.
Fair enough, thanks for explaining.
create, upload, download, diff, and lint projects directly from Github - closes #169.
Inputs starting with
git+(e.g. git+/path/to/repo.git) with will have thegit+stripped and passed directly to thegitcommand-line. Alternatively URLs starting withgit:will be passed as is to thegitcommand-line. This seems to roughly correspond topip.