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

~bzr-pqm/bzr/bzr.dev » Changes from revision 0.17.45

Rev	Summary	Authors	Date
0.17.45	Just make sure we have the right decompressor Just make sure we have the right decompressor	John Arbash Meinel	15 years ago
0.17.44	Use the bit field to allow both lzma groups and zlib groups. Use the bit field to allow both lzma groups and zlib groups. This lets 'bzr pack' decide which format to use. Helpful for testing.	John Arbash Meinel	15 years ago
0.17.43	Give better chk and inventory progress indication during pac... Give better chk and inventory progress indication during pack/autopack	John Arbash Meinel	15 years ago
0.17.42	Handle bzr.dev changes. Handle bzr.dev changes.	Robert Collins	15 years ago
0.17.41	brisbane core has removed InterModel, so always set rich_roo... brisbane core has removed InterModel, so always set rich_root=True	John Arbash Meinel	15 years ago
0.17.40	CHKSerializer no longer has a 'parent_id_basename_index' fie... CHKSerializer no longer has a 'parent_id_basename_index' field to care about.	John Arbash Meinel	15 years ago
0.17.39	All results are bytes now. All results are bytes now.	Robert Collins	15 years ago
0.17.38	fix registration of hash16chk+gc fix registration of hash16chk+gc	Ian Clatworthy	15 years ago
0.17.37	fix initialization of start variable fix initialization of start variable	Ian Clatworthy	15 years ago
0.17.36	Adding a mini-len to the delta/fulltext bytes Adding a mini-len to the delta/fulltext bytes This adds 1.6bytes/entry for inventory pages, and 2.5 bytes/entry for text pages. But that is down in the <1% increased content. The main advantage is that if we get rid of labels, it allows the content to be fully self describing. Especially beneficial for chk pages, as you could regenerate the chk index from the .pack file, without including any labels. (the label is the sha1 sum). It does increase the time to extract, as you are now decoding those little bytes in the beginning. However, that time could be significantly decreased with a Pyrex decoder. (at the moment the overhead is 0.2s out of 14s under lsprof.) If we decide to go with labels, then the 1% is a bit superflous, but if we get rid of labels, we trade 1% here for 40% in the labels.	John Arbash Meinel	15 years ago
0.17.35	Get the tests passing again Get the tests passing again Also fix the Compressor.extract() for the new changes.	John Arbash Meinel	15 years ago
0.17.34	Another disk-format bump. Another disk-format bump. Move the labels/sha1 information into a pre-header. This also makes it easier to decide to enable/disable the headers, as we can support both with the same deserialising code (at least until we remove the extra info from the indexes.) This also makes a fulltext record stream start with 'f' and a delta record stream start with 'd', which makes them more self describing. The next step would probably be to write the base128 length of the encoded bytes, which would make them fully independent, though you wouldn't know what content they refer to. This also brings in an update to .compress() which allows us to see that we overflowed our group, roll back and start a new one. This seems to give better compression in a 'more stable' manner. Still open to tweaking, though. Also introduce the 'gcc-chk255-big' which uses 64k leaf pages rather than 4k leaf pages. Initial results show smaller compressed size at a small (10%) increase in uncompressed size. Also shows a full level decrease in the tree depth. No-labels decreases the inv size approx 300kB, and big-page decreases the inv size another 300kB, not to mention the 116k decrease in the .cix index, just from not having the extra pages. Having both no-labels and big inv pages brings a total drop of 11023k down to 9847k for the repo (1176kB savings, or 10% overall). For now, leave the default with labels, but consider changing it.	John Arbash Meinel	15 years ago
0.17.33	A couple tweaks to repofmt to allow fetching to work again. A couple tweaks to repofmt to allow fetching to work again.	John Arbash Meinel	15 years ago
0.17.32	Remove the equivalence table tests, since we don't use it an... Remove the equivalence table tests, since we don't use it anymore.	John Arbash Meinel	15 years ago
0.17.31	Bring in the 'rabin' experiment. Bring in the 'rabin' experiment. Change the names and disk-strings for the various repository formats. Make the CHK format repositories all 'rich-root' we can introduce non-rich-root later. Make a couple other small tweaks, like copyright statements, etc. Remove patch-delta.c, at this point, it was only a reference implementation, as we have fully integrated the patching into pyrex, to allow nicer exception handling.	John Arbash Meinel	15 years ago
0.17.30	Merge trunk Merge trunk	Robert Collins	15 years ago
0.17.29	Update for VersionedFiles adapter API change. Update for VersionedFiles adapter API change.	Robert Collins	15 years ago
0.17.28	Do not register a chk repository format unless chk is availa... Do not register a chk repository format unless chk is available.	Robert Collins	15 years ago
0.17.27	Fix typo. Fix typo.	Robert Collins	15 years ago
0.17.26	Working better --gc-plain-chk. Working better --gc-plain-chk.	Robert Collins	15 years ago

Older »