lp:~mwhudson/loggerhead/less-stupid-whole-history-redundancy

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:loggerhead

Paul Hummer (community): Approve on 2009-04-24

Loggerhead Team: Pending requested 2009-04-24

Diff: None lines api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

338. By Michael Hudson-Doyle on 2009-04-24

oh heh, _full_history wasn't what I thought it was

337. By Michael Hudson-Doyle on 2009-04-24

don't cache _full_history either, which doesn't make much difference really...

336. By Michael Hudson-Doyle on 2009-04-24

probably daft optimization

335. By Michael Hudson-Doyle on 2009-04-24

don't cache _rev_indices either

334. By Michael Hudson-Doyle on 2009-04-24

it takes < 0.1 second to derive _revno_revid from the _rev_info, so don't cache
it!

333. By Michael Hudson-Doyle on 2009-04-24

eliminate some of the more egregious memory usage.
reduces whole history size by another 14% for launchpad branches

332. By Michael Hudson-Doyle on 2009-04-24

score one for the memory profiling we did at pycon: store the "where merged"
information in small tuples, not small sets.
this makes the 'whole history' cache for a launchpad branch take up about 15%
less RAM.

331. By Michael Hudson-Doyle on 2009-04-23

fix bug #358322: only create one temporary sql dir per process (Matt Nordhoff)

330. By Michael Hudson-Doyle on 2009-04-23

merge Matt's one-config branch, which starts to use a single object for configuration

329. By Matt Nordhoff on 2009-04-17

Don't use the "sets" module.

1.) The code wasn't using the built-in "set" type when it was available.

2.) bzrlib isn't compatible with Python <2.4 anyway, so there's no need to fall back to the "sets" module.