At http://bzr.arbash-meinel.com/branches/bzr/brisbane/hack

------------------------------------------------------------
revno: 3837
revision-id: [email protected]
parent: [email protected]
parent: [email protected]
committer: John Arbash Meinel <[email protected]>
branch nick: hack
timestamp: Wed 2009-01-07 16:22:49 -0600
message:
  Merge in latest version of brisbane-core.
modified:
  bzrlib/chk_map.py              chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
  bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
    ------------------------------------------------------------
    revno: 3802.3.3
    revision-id: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: brisbane-core
    timestamp: Wed 2009-01-07 16:15:57 -0600
    message:
      Change deserialise to properly handle when there is a '\r' in the key.
      
      It turns out that bytes.splitlines() can split on a plain '\r', which 
breaks things.
      So instead, we use bytes.split('\n') which work correctly, though we have 
to remove
      the extra '' at the end.
    modified:
      bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
      bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
    ------------------------------------------------------------
    revno: 3802.3.2
    revision-id: [email protected]
    parent: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: brisbane-core
    timestamp: Wed 2009-01-07 15:58:28 -0600
    message:
      Bring in the common-prefix-extraction code.
      
      This also updates a few tests to be less dependent on the exact 
serialization.
      It also changes the disk signature, since repositories created with
      the new chk map pages are incompatible with existing repositories.
      
      If we really need it, we could parameterize the prefix extraction portion,
      but it doesn't seem worthwhile at this point.
    modified:
      bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
      bzrlib/repofmt/pack_repo.py    
pack_repo.py-20070813041115-gjv5ma7ktfqwsjgn-1
      bzrlib/repository.py           
rev_storage.py-20051111201905-119e9401e46257e3
      bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.16
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 15:54:42 -0600
        message:
          Split out the unmap_pages_in tests and add one that shows if we
          already have enough items loaded in memory, it doesn't page in 
anything else.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.15
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 15:08:19 -0600
        message:
          Update the unmap_pages_in to also handle specific page-cache edge 
cases.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.14
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:56:10 -0600
        message:
          Change the deep_split code to use a bigger value so we still get a 
split.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.13
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:53:33 -0600
        message:
          We were incorrectly using '' when we wanted to use None as the common 
prefix.
        modified:
          bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.12
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:42:28 -0600
        message:
          Fix up the LeafNode direct tests, and add some tests with serialized
          forms that include a common prefix.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.11
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:35:40 -0600
        message:
          Change the multiple_maps test to also be independent of sha hash.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.10
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:28:50 -0600
        message:
          Change the IterInteresting to avoid sha hashes and be serialization 
agnostic.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.9
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 14:25:44 -0600
        message:
          Fix the IterInteresting tests.
          
          It would be nice to make them independent of serialization, though.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.8
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:56:19 -0600
        message:
          Fix the other k1_k22_k23 test
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.7
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:53:40 -0600
        message:
          Fix up the k1_from_k1 test.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.6
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:47:00 -0600
        message:
          Update a bunch of tests that were using None rather than not 
displaying keys.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.5
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:44:35 -0600
        message:
          Properly fix up the dump_tree tests, we now suppress the keys by 
default.
        modified:
          bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.4
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:38:03 -0600
        message:
          Change child_child to use _dump_tree,
          update _dump_tree to allow suppressing the node keys.
          Since it often doesn't matter for the tests.
        modified:
          bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.3
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:18:12 -0600
        message:
          Change the apply_ab_empty and apply_empty_ab tests to make them 
slightly more sha1 independent.
          
          They generally are testing the serialized form, and not the actual 
key.
        modified:
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.2
        revision-id: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 13:10:21 -0600
        message:
          Change the term 'lookup' to the term 'search', as it is closer to 
what Robert envisioned.
        modified:
          bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
        ------------------------------------------------------------
        revno: 3813.1.1
        revision-id: [email protected]
        parent: [email protected]
        parent: [email protected]
        committer: John Arbash Meinel <[email protected]>
        branch nick: prefix
        timestamp: Wed 2009-01-07 12:34:04 -0600
        message:
          Bring in the tip of brisbane-core, including the remap fixes.
        modified:
          bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
          bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
    ------------------------------------------------------------
    revno: 3802.3.1
    revision-id: [email protected]
    parent: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: brisbane-core
    timestamp: Wed 2009-01-07 12:08:39 -0600
    message:
      Change _iter_nodes so that it is a generator.
      
      Change _check_remap to use _iter_nodes to eliminate the redundancy.
      We also added a 'batch_size' parameter, so that with large fan-outs
      we don't page in all the nodes unnecessarily. Which adds both direct
      computation overhead, but also causes us to thrash the page cache.
    modified:
      bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
      bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
    ------------------------------------------------------------
    revno: 3802.2.7
    revision-id: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: remap
    timestamp: Wed 2009-01-07 12:05:53 -0600
    message:
      Change _iter_nodes into a generator.
      
      This dramatically simplifies _check_remap, because all of the code to
      handle paging in new nodes is already present.
      
      All we needed to add was the ability to 'batch' requests for the
      get_record_stream(), instead of reading in everything at once.
    modified:
      bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
    ------------------------------------------------------------
    revno: 3802.2.6
    revision-id: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: remap
    timestamp: Wed 2009-01-07 11:54:32 -0600
    message:
      fix a test case that exercises the new stop-early code.
    modified:
      bzrlib/tests/test_chk_map.py   
test_chk_map.py-20081001014447-ue6kkuhofvdecvxa-2
    ------------------------------------------------------------
    revno: 3802.2.5
    revision-id: [email protected]
    parent: [email protected]
    committer: John Arbash Meinel <[email protected]>
    branch nick: remap
    timestamp: Wed 2009-01-07 11:52:52 -0600
    message:
      Change _check_remap to only page in a batch of children at a time.
      
      If we have more than 16 children, it is unlikely that we will be able to 
fit
      everything into a single node anyway, and it helps prevent us from 
requesting
      all 255 nodes and overflowing the page cache.
    modified:
      bzrlib/chk_map.py              
chk_map.py-20081001014447-ue6kkuhofvdecvxa-1
Diff too large for email (1726 lines, the limit is 1000).
-- 
bazaar-commits mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/bazaar-commits

Reply via email to