Merge lp:~mbp/bzr/192859-2.0-symlinks into lp:bzr/2.0
Proposed by
Martin Pool
Status: | Merged | ||||||||
---|---|---|---|---|---|---|---|---|---|
Approved by: | Martin Pool | ||||||||
Approved revision: | no longer in the source branch. | ||||||||
Merged at revision: | 4756 | ||||||||
Proposed branch: | lp:~mbp/bzr/192859-2.0-symlinks | ||||||||
Merge into: | lp:bzr/2.0 | ||||||||
Diff against target: |
67 lines (+25/-12) 1 file modified
bzrlib/tests/per_workingtree/test_symlinks.py (+25/-12) |
||||||||
To merge this branch: | bzr merge lp:~mbp/bzr/192859-2.0-symlinks | ||||||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
John A Meinel | Needs Fixing | ||
Review via email: mp+30180@code.launchpad.net |
Commit message
cope with links changing into directories
Description of the change
This fixes another symlink problem case: adding a file under a directory that was a symlink in the basis revision.
The really right fix for this is to get away from wt inventory entries having a defined in-memory kind: their kind is whatever is on disk. But doing that would probably be much too big for 2.0. I might try a 2.3 branch for it later.
There may be more problem cases.
To post a comment you must log in.
My only comment is that the tests for "tree.smart_ add(['tree/ a/file' ])", etc, don't actually assert the state of the final tree. Just that tree.smart_add() doesn't fail.
Which I think does prove that bug 192859 is fixed, but it would be more useful as an ongoing test to not just assert that the functions don't fail with an exception, but that they did have the expected effect (the tree now sees a dir, etc.)