diff: use RepoTree to compare directory contents #4518
Conversation
pmrowla
moved this from In progress
to Review in progress
in DVC 25 August - 8 September 2020
dvc/repo/diff.py
Outdated
| # if dir hash is missing from cache, and no remote to pull it from, | ||
| # there is nothing we can do here |
There was a problem hiding this comment.
Should we print a warning or at least a debug message?
|
|
||
|
|
||
| def _output_paths(repo): | ||
| on_working_tree = isinstance(repo.tree, LocalTree) |
There was a problem hiding this comment.
Does on_working_tree matter anymore though? RepoTree will just use repo.tree and will act accordingly if it is a wtree or a git rev tree.
There was a problem hiding this comment.
So the issue here is that DvcTree (and RepoTree) does not re-compute dir output hashes when the workspace is dirty. I'm not sure that we would want to change that behavior just for dvc diff, it seems like in a majority of scenarios we would want to re-use the already computed hashes for dir outputs when we are using DvcTree and RepoTree.
There was a problem hiding this comment.
That's actually an inconsistency that should be solved in RepoTree. Ideally, we should be able to work with both dirty state and clean repos. But could start with the current approach too.
There was a problem hiding this comment.
Filed #4523 for this issue
pmrowla
force-pushed
the
diff-repo-tree
branch
from
5d40d76
to
7660513
Compare
❗ I have followed the Contributing to DVC checklist.
📖 If this PR requires documentation updates, I have created a separate PR (or issue, at least) in dvc.org and linked it here.
Thank you for the contribution - we'll try to review it as soon as possible. 🙏
Partially addresses #2982.
dir/added/modified/deleted status)RepoTree(stream=True)This does not implement pulling all uncached file contents by default, or the file size related tasks.