Merge lp:~gz/bzr-git/import_os_not_posix_841177 into lp:bzr-git
Proposed by
Martin Packman
Status: | Merged | ||||
---|---|---|---|---|---|
Merged at revision: | 1357 | ||||
Proposed branch: | lp:~gz/bzr-git/import_os_not_posix_841177 | ||||
Merge into: | lp:bzr-git | ||||
Diff against target: |
53 lines (+18/-0) (has conflicts) 1 file modified
workingtree.py (+18/-0) Text conflict in workingtree.py |
||||
To merge this branch: | bzr merge lp:~gz/bzr-git/import_os_not_posix_841177 | ||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
bzr-git developers | Pending | ||
Review via email: mp+74023@code.launchpad.net |
Description of the change
Makes bzr-git stick to the `os` module, which is a platform independent wrapper for the builtin `posix`, `nt` and so on modules. There is no behaviour change here, apart from making the module importable on non-posix systems.
I resisted changing `posixpath` usage into `os.path` because it's not trivial to tell which is actually desired. Paths from the filesystem should be operated on with `os.path` but normalized bzrlib paths that use forward slashes only can use most of the posixpath functions safely.
To post a comment you must log in.
To answer a query from IRC, stat_result doesn't have different content or behaviour in the cases here. See statresult_new in Modules/ posixmodules. c which is pretty straight forward. (The module source confusingly has 'posix' in the name but is cross-platform code, and results in a module named after the platform).
The conflicts with trunk are a little crazy, but should be easy enough to resolve correctly now.