Permalink
Commits on Apr 4, 2016
  1. Merge pull request #30 from LennyLinux/patch-1

    update README to reflect btree package change
    committed Apr 4, 2016
Commits on Apr 1, 2016
Commits on Mar 30, 2016
  1. Merge pull request #29 from peterbourgon/google-btree

    Switch from petar/GoLLRB to google/btree
    committed Mar 30, 2016
  2. Switch from petar/GoLLRB to google/btree

    Closes issue #28
    committed Mar 30, 2016
Commits on Nov 13, 2015
  1. Merge pull request #26 from percona/master

    Return real error from Erase()
    committed Nov 13, 2015
Commits on Sep 5, 2015
Commits on Apr 21, 2015
Commits on Dec 31, 2014
  1. Merge pull request #22 from gprggr/diskv

    diskv: missing f.Close() in writeStreamWithLock(...)
    committed Dec 31, 2014
Commits on Dec 30, 2014
Commits on Dec 12, 2014
  1. Merge pull request #21 from jonboulle/nocache

    ReadStream: never siphon if CacheMaxSize <= 0
    committed Dec 12, 2014
  2. ReadStream: never siphon if CacheMaxSize <= 0

    If the user has not set a CacheMaxSize >0, we should never attempt to
    start siphoning values while reading objects.
    jonboulle committed Dec 12, 2014
  3. Merge pull request #19 from jonboulle/rename

    Import: don't fail for cross-device imports
    committed Dec 12, 2014
  4. Merge pull request #20 from jonboulle/panic

    KeysPrefix: do not transform empty prefix
    committed Dec 12, 2014
  5. KeysPrefix: do not transform empty prefix

    When KeysPrefix was introduced, the behaviour of Keys was changed:
    rather than just walking from BasePath, it attempted to transform the
    empty prefix using the configured BlockTransform. I do not think this is
    correct behaviour; it just happened to be fine in the default case
    because the default transform used in tests would simply return an empty
    slice, which means pathFor() would end up returning BasePath anyway.
    However, BlockTransform functions should arguably not be expected to
    successfully transform an empty key into an empty slice (ours actually
    panics in that scenario which is how I noticed this ;-)
    jonboulle committed Dec 12, 2014
  6. Import: don't fail for cross-device imports

    rename(2) only works on a single device. When trying to import a file
    (with move=true) from a different device to the one on which the diskv
    is located, we should gracefully fall back to copying the file and then
    removing it (like mv(1) does).
    jonboulle committed Dec 12, 2014
  7. Merge pull request #18 from peterbourgon/import

    Add Import method
    committed Dec 12, 2014
  8. Merge branch 'master' into import

    Conflicts:
    	diskv.go
    committed Dec 12, 2014
  9. Code cleanliness updates

    - Factor out cache busting code
    - Make explicit WithLock, WithRLock in function signatures
    committed Dec 12, 2014
Commits on Dec 11, 2014
  1. Add Import

    We don't need to worry about caching.
    That happens on read.
    committed Dec 11, 2014
  2. Ensure ReadStream with direct=true isn't racy

    Partially addresses #17. Big thanks to @jonboulle for spotting that
    head-slapper.
    committed Dec 11, 2014
Commits on Dec 9, 2014
  1. Merge pull request #15 from peterbourgon/cancel-keys-walk

    Add cancel parameter to Keys[Prefix]
    committed Dec 9, 2014
  2. Fix test consistency

    committed Dec 9, 2014
  3. Update CAS example

    committed Dec 9, 2014
Commits on Dec 8, 2014
  1. Minor code cleanup

    - Uniformly filepath instead of path
    - Use consistent declaration style
    - Improve variable naming conventions re: ADG's Paris talk
    - Avoid stuttering in some struct member naming
    committed Dec 8, 2014
  2. Merge pull request #12 from peterbourgon/keysprefix

    Add KeysPrefix method
    committed Dec 8, 2014
Commits on Dec 6, 2014
  1. Add KeysPrefix method

    Addresses #11.
    committed Dec 6, 2014
Commits on Jul 9, 2014
Commits on Jul 7, 2014
  1. Merge pull request #10 from peterbourgon/read-direct

    Add direct flag to ReadStream
    committed Jul 7, 2014