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

~bzr-pqm/bzr/bzr.dev » Changes from revision 4593.4.9

From Revision 4593.4.9 to 4583

Rev	Summary	Authors	Date
4593.4.9	Add some debugging statements for now. Add some debugging statements for now. It seems that using get_ancestry in a real-world scenario is an even bigger improvement. I see a 5x improvement in getting the ancestry of bzr.dev. It is ~300ms instead of being 1.5s. Time to try testing it on an OOo tree.	John Arbash Meinel	15 years ago
4593.4.8	Implement CombinedGraphIndex.get_ancestry() Implement CombinedGraphIndex.get_ancestry() This is the appropriate wrapper around GraphIndex/BtreeIndex.get_ancestry. It should allow quick access to getting the complete ancestry graph of the given set of keys and do so in a fairly efficient fashion.	John Arbash Meinel	15 years ago
4593.4.7	Basic implementation of a conforming interface for GraphInde... Basic implementation of a conforming interface for GraphIndex. Add some simple tests, for now we won't try to do anything fancier. Time to start working on CombinedGraphIndex's interface.	John Arbash Meinel	15 years ago
4593.4.6	Add a test that we can walk some of the keys on a given page Add a test that we can walk some of the keys on a given page and that we stop at the boundary, but we could resume walking from there.	John Arbash Meinel	15 years ago
4593.4.5	Start adding some tests. Start adding some tests. Extend the code a little bit, to make parent_map and missing_keys be state that is passed in and then mutated, rather than one and not the other. We need the state since we are walking more than the minimal, though passing in the 'goal' of the function does seem a little silly.	John Arbash Meinel	15 years ago
4593.4.4	Trying out a few more tweaks. Trying out a few more tweaks.	John Arbash Meinel	15 years ago
4593.4.3	Some minor attribute lookup cleanus, doesn't make a big diff... Some minor attribute lookup cleanus, doesn't make a big difference.	John Arbash Meinel	15 years ago
4593.4.2	Removing the min(keys) and max(keys) calls saves 100ms in th... Removing the min(keys) and max(keys) calls saves 100ms in the inner loop (get_ancestry() over all of bzr.dev in a single index is 347ms => 245ms). The current breakdown is roughly: 0.4789 0.1127 bzrlib.btree_index:1129(get_ancestry) 0.0418 0.0418 +<method 'update' of 'set' objects> 0.0480 0.0325 +bzrlib.btree_index:966(_multi_bisect_right) 0.0274 0.0274 +<method 'difference' of 'set' objects> 0.0081 0.0081 +<method 'add' of 'set' objects> 0.0075 0.0075 +<sorted> 0.0048 0.0004 +bzrlib.btree_index:899(_get_internal_nodes) 0.2275 0.0004 +bzrlib.btree_index:917(_get_leaf_nodes) 0.0002 0.0002 +<method 'extend' of 'list' objects> 0.0009 0.0001 +bzrlib.btree_index:1375(key_count) So we have a bit of just general overhead (112ms), and then 50ms spent in _multi_bisect_right, which we could move to a C extension. 50ms in set.update and 28ms in set.difference 227ms in reading and parsing the 222 nodes from disk. It seems a little unfortunate that parsing is the primary overhead, but previous investigation did not reveal much fat that could be trimmed. It is 3.8MB of uncompressed data that is being parsed. That's got to take some amount of time. 200ms might be reasonable. Which would hint that the only way to speed it up would be: 1) a different format 2) don't read the whole thing, stupid :)	John Arbash Meinel	15 years ago
4593.4.1	Implement a function on btree that inlines the get_parent_ma... Implement a function on btree that inlines the get_parent_map loop. It basically looks for direct parents within a single LeafNode, until it hits one that goes outside that node. Also include some performance notes. At least for the number of calls, the new function is much better than get_parent_map. It takes only 50 calls to walk all of bzr.dev, while for get_parent_map it takes 1108 calls. Part of that is the 'long tail problem', which is better, albeit not perfect with the new function.	John Arbash Meinel	15 years ago
4593	(robertc) Partial overhaul of check to do less duplicate wor... (robertc) Partial overhaul of check to do less duplicate work. (Robert Collins)	Canonical.com Patch ...	15 years ago
4592	Updates to buildout.cfg etc to have 'make installer-all' sta... Updates to buildout.cfg etc to have 'make installer-all' start being the preferred way to build win32 installer.	Canonical.com Patch ...	15 years ago
4591	(jam) Small 'const' fixes for diff-delta.c (bug #408441) (jam) Small 'const' fixes for diff-delta.c (bug #408441)	Canonical.com Patch ...	15 years ago
4590	(Jelmer) Document bzr_transports property in the plugin API. (Jelmer) Document bzr_transports property in the plugin API.	Canonical.com Patch ...	15 years ago
4589	(bialix) Include the image format plugins in the win32 all-i... (bialix) Include the image format plugins in the win32 all-in-one installer.	Canonical.com Patch ...	15 years ago
4588	(jam) Get bundles working with --2a (bug #393349) (jam) Get bundles working with --2a (bug #393349)	Canonical.com Patch ...	15 years ago
4587	(mbp) further LockableFiles cleanups (mbp) further LockableFiles cleanups	Canonical.com Patch ...	15 years ago
4586	(mbp) updates to progress-related docstrings and remove anot... (mbp) updates to progress-related docstrings and remove another obsolete method	Canonical.com Patch ...	15 years ago
4585	(mbp) KnownFailure for bug 408193 (mbp) KnownFailure for bug 408193	Canonical.com Patch ...	15 years ago
4584	(mbp) fix selftest progress and error display problems (mbp) fix selftest progress and error display problems	Canonical.com Patch ...	15 years ago
4583	(mbp) Cleanup TestUIFactory; prevent pb.tick() failures (mbp) Cleanup TestUIFactory; prevent pb.tick() failures	Canonical.com Patch ...	15 years ago

Older »