~bzr-pqm/bzr/bzr.dev : changes

~bzr-pqm/bzr/bzr.dev » Changes from revision 3514.4.37

Rev	Summary	Authors	Date
3514.4.37	Tests pass when using _entries_lca as the default algorithm. Tests pass when using _entries_lca as the default algorithm. The only test that doesn't pass is a bug in MultiWalker, which has a fix being merged already.	John Arbash Meinel	16 years ago
3514.4.36	Optimizing how we extracted multiple trees left a path that ... Optimizing how we extracted multiple trees left a path that didn't create a base_tree. Fixed.	John Arbash Meinel	16 years ago
3514.4.35	Add a test that we only call get_symlink_target if the objec... Add a test that we only call get_symlink_target if the object should be a symlink.	John Arbash Meinel	16 years ago
3514.4.34	Handle symlinks properly when objects are not RevisionTrees Handle symlinks properly when objects are not RevisionTrees	John Arbash Meinel	16 years ago
3514.4.33	Implement support for 'tree-reference'. Implement support for 'tree-reference'. It seems that _entries3 always returns False for content changed, I would have thought it would look at entry.revision in those circumstances. Anyway, it seems to work for what we need.	John Arbash Meinel	16 years ago
3514.4.32	Add a test for proper behavior when everything is a WT. Add a test for proper behavior when everything is a WT. This only really happens in the test suite, you'll never get a LCA Tree as a WT in standard behavior (I don't know if you could get a BASE tree as a wT). Also, clean up some trailing whitespace.	John Arbash Meinel	16 years ago
3514.4.31	Add expected failures for cases where we should be looking a... Add expected failures for cases where we should be looking at more than just BASE + LCAs, such as intermediate LCAs, or the whole per-file/attribute graph. They are fairly unlikely, but it is good to know that we don't handle them yet.	John Arbash Meinel	16 years ago
3514.4.30	Several updates. Several updates. Add a document describing the merge algorithm, and the rationale for the different steps. Change the sha1 algorithm to use get_file_sha1() rather than assuming the inventory entries text_sha1 value is correct. (It is always None for working trees.) For sha1-values specifically, don't allow overruling an LCA value when one tip seems to supersede the other's value. We will still do this for names and parent_ids. Add tests that we conflict if lcas disagree and tips do not have the same content, even if one of them is an lca value, and one is newer.	John Arbash Meinel	16 years ago
3514.4.29	We have added the symlink tests, and now have covered all th... We have added the symlink tests, and now have covered all the ones I was aware of.	John Arbash Meinel	16 years ago
3514.4.28	More symlink tests. More symlink tests. Test that we properly track content changes for symlinks, and that we merge them correctly. Also add a test for the expected failure of the 3-way merge logic on content of symlinks (and files.)	John Arbash Meinel	16 years ago
3514.4.27	A couple of symlink tests, we need to do more. A couple of symlink tests, we need to do more.	John Arbash Meinel	16 years ago
3514.4.26	Clean up comments, only symlink support left. Clean up comments, only symlink support left.	John Arbash Meinel	16 years ago
3514.4.25	A few more merge-level behavior tests. A few more merge-level behavior tests.	John Arbash Meinel	16 years ago
3514.4.24	Implement support for 'interesting_files' and 'interesting_i... Implement support for 'interesting_files' and 'interesting_ids' for _entries_lca	John Arbash Meinel	16 years ago
3514.4.23	Handle when there are more than 2 LCAs while searching for t... Handle when there are more than 2 LCAs while searching for the unique lca.	John Arbash Meinel	16 years ago
3514.4.22	Handle executable bit changes as well. Handle executable bit changes as well.	John Arbash Meinel	16 years ago
3514.4.21	Hook up the lca-way logic into the path handlers. Hook up the lca-way logic into the path handlers. Add at least 1 test which would fail without using this logic. Lots more tests to follow. Mostly stuff that we can't really test at the lower levels.	John Arbash Meinel	16 years ago
3514.4.20	Use the _lca_multi_way to work out if there is actually a ki... Use the _lca_multi_way to work out if there is actually a kind/parent/name/content change. This lets us filter out nodes that aren't interesting, as well as properly set the 'content_changed' flag.	John Arbash Meinel	16 years ago
3514.4.19	Add the _lca_multi_way function, and explicit tests. Add the _lca_multi_way function, and explicit tests. There may be some other edge cases, but I think I covered them all.	John Arbash Meinel	16 years ago
3514.4.18	Bring in both branch-builder threads. Bring in both branch-builder threads.	John Arbash Meinel	16 years ago

Older »