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

~bzr-pqm/bzr/bzr.dev » Changes from revision 4343.3.11

Rev	Summary	Authors	Date
4343.3.11	Change PendingAncestryResult to strip ghosts from .get_keys(... Change PendingAncestryResult to strip ghosts from .get_keys()	John Arbash Meinel	15 years ago
4343.3.10	Add a per_repository_reference test with real stacked repos. Add a per_repository_reference test with real stacked repos. This adds a test that sets up a stacked repository with an extra revision versus the stacked-on location, and then asserts you can branch from this. I would think this would be tested somewhere, but it wasn't a per-repository-reference test, so when I broke the code, it didn't fail an obvious test. This now changes the code so that if there are no keys to be fetched for the 'missing_keys', then it doesn't try to get a stream from that vf. There wasn't any real reason to do so, and it avoids trying to get a stream for chk_bytes when the vf doesn't exist.	John Arbash Meinel	15 years ago
4343.3.9	Remove an accidental 'bork'. Remove an accidental 'bork'. Though it does show I was correct about the code not being used.	John Arbash Meinel	15 years ago
4343.3.8	Some cleanup passes. Some cleanup passes. move test_resume_chk_bytes from per_repository_referece to test_pack_repository, as it is a bit more implementation specific. Make sure we don't allow get_stream_for_missing_keys() if we have a pending stream.	John Arbash Meinel	15 years ago
4343.3.7	Update the suspend/resume/commit/abort_write_group tests for... Update the suspend/resume/commit/abort_write_group tests for CHK1. It turns out that the parameterized tests weren't actually parameterized, always only using the default format. Added a test that 'commit_write_group()' also doesn't leave any detritus around in upload, and puts things into packs and indices correctly. Updated the 'suspend' test to test that all indices are present, and not just that the .pack file is. Updated the ResumedPack.finish() and ResumedPack.abort() to handle .cix entries.	John Arbash Meinel	15 years ago
4343.3.6	Take out the _delta_across_repos flag. Take out the _delta_across_repos flag. Instead, we now check that the either commit_write_group fails, or get_record_stream(..., True) succeeds. Also, change the make_source_with_delta_record so that it generates the closest thing to a delta for chk formats. We have to use 'insert_record_stream', because 'add_lines' doesn't delta for chk formats.	John Arbash Meinel	15 years ago
4343.3.5	small comment small comment	John Arbash Meinel	15 years ago
4343.3.4	Custom implementation for GroupCHKStreamSource.get_stream_fo... Custom implementation for GroupCHKStreamSource.get_stream_for_missing_keys. This now ensures that all referenced chk pages will be transmitted when we are requested to send missing inventories. Also, adding a file to use for testing 'chk_bytes' specific information when testing stacking.	John Arbash Meinel	15 years ago
4343.3.3	Be sure to check for missing compression parents in chk_byte... Be sure to check for missing compression parents in chk_bytes	John Arbash Meinel	15 years ago
4343.3.2	All stacking tests seem to be passing for dev6 repos All stacking tests seem to be passing for dev6 repos 1) Add _GCGraphIndex.scan_unvalidated_index() It currently is a no-op because we a) Don't cache the possible external parent references b) Don't have external compression parents, so we can't have missing compression parents that we need to scan either. 2) Introduce a .resumed_pack_factory and ResumedGCPack We only need this to handle the fact that '_get_external_refs' is always empty. It would probably be nicer if we could do this with a different flag.. 3) RepositoryFormat._deltas_across_repos I needed a way to disable the 'cannot commit because we have a delta' tests for GC repositories, because they currently have no way to represent a delta between pack files. So I added this flag for now. 4) ResumedPack now properly supports chk_index, and we add a test to ensure this fact.	John Arbash Meinel	15 years ago
4343.3.1	Set 'supports_external_lookups=True' for dev6 repositories. Set 'supports_external_lookups=True' for dev6 repositories. I'm choosing not to do a new format, since it is dev anyway... At the moment, that means supporting 'suspend_write_group', and implementing '_index.get_missing_parents'.	John Arbash Meinel	15 years ago
4343	(Johan Walles) fix bug #180116 by using a sort() and linear ... (Johan Walles) fix bug #180116 by using a sort() and linear operation for osutils.minimum_path_selection()	Canonical.com Patch ...	15 years ago
4342	(robertc) Do not trigger NoSuchRevision pulling from a stack... (robertc) Do not trigger NoSuchRevision pulling from a stacked branch when the client needs content from it and the stacked-on branch. (Robert Collins)	Canonical.com Patch ...	15 years ago
4341	(Jelmer) Mention --force in the error message when unable to... (Jelmer) Mention --force in the error message when unable to access in the master branch in 'bzr switch'.	Canonical.com Patch ...	15 years ago
4340	(Jelmer) Add InterBranch.pull. (Jelmer) Add InterBranch.pull.	Canonical.com Patch ...	15 years ago
4339	(igc) reconfigure --standalone no longer raises (igc) reconfigure --standalone no longer raises IncompatibleRepositories (Martin von Gagern)	Canonical.com Patch ...	15 years ago
4338	Revert previous patch, Revert previous patch, PQM now presents a unicode friendly file system encoding	Canonical.com Patch ...	15 years ago
4337	testing pqm unicode support testing pqm unicode support	Canonical.com Patch ...	15 years ago
4336	(igc) pull --local (Gary van der Merwe) (igc) pull --local (Gary van der Merwe)	Canonical.com Patch ...	15 years ago
4335	(igc) Raise an error for unknown eol values. (Brian de Alwis... (igc) Raise an error for unknown eol values. (Brian de Alwis)	Canonical.com Patch ...	15 years ago

Older »