Skip to content
This repository

Apr 15, 2014

  1. Jens Alfke

    Revert "CBLT-268: product version now set to a token, instantiated at…

    … build time:"
    
    This reverts commit ee5212d.
    authored April 15, 2014

Apr 12, 2014

  1. Jens Alfke

    Bumped version number to 1.5 for GA

    authored April 11, 2014

Apr 11, 2014

  1. Jens Alfke

    Catch uncaught exceptions in CBLDatabase doAsync etc.

    Otherwise GCD tends to kill the process, which means unit testing can’t complete.
    Fixes #309
    authored April 11, 2014
  2. Jens Alfke

    Added some missing copyright/license comments.

    authored April 11, 2014
  3. Jens Alfke

    API Change: Added CBLAuthenticator

    Pre-existing CBLReplication auth methods are still present,
    but the ones supported by CBLAuthenticator are deprecated.
    REST API replication auth works as before.
    authored April 11, 2014

Apr 07, 2014

  1. Jens Alfke

    Added a unit test verifying revpos changes when attachment is updated.

    authored April 07, 2014

Apr 03, 2014

  1. Jens Alfke

    Safer handling of unexpected revisions returned from _bulk_get

    Don’t process such a revision; it’ll cause an exception later.
    For #300
    authored April 03, 2014

Apr 02, 2014

  1. Jens Alfke

    Fix replicator checkpoint issue by making sure checkpoint doc ID stay…

    …s constant
    
    Memoize the value returned by .remoteCheckpointDocID so it stays the same even
    if a setting changes — namely, the -stop method sets _continuous=NO for some reason.
    Fixes #301
    authored April 02, 2014
  2. Jens Alfke

    Removed support for “_conflicts” property in doc passed to map blocks.

    It didn’t work quite right (could miss some conflicts) and taking it out speeds up
    view indexing a little bit.
    Developers who want to find conflicts should use an allDocs query with the
    only-conflicts option instead.
    Fixes #159
    authored April 02, 2014

Apr 01, 2014

  1. Jens Alfke

    Merge pull request #302 from couchbase/delete_conflict_test

    Unit test regarding conflict resolution behavior
    authored April 01, 2014

Mar 31, 2014

  1. Jens Alfke

    Removed “Slow comparison” log message from JSON collator

    This wasn’t supposed to be checked in…
    authored March 31, 2014
  2. Jens Alfke

    Reduced max memory usage of push replicator

    - Added some fixes to avoid keeping redundant data in memory
    - Constrained max size of a _bulk_docs request to keep the object tree from getting too large
    Fixes #270
    authored March 31, 2014

Mar 27, 2014

  1. Jens Alfke

    Fixed -[CBLDatabase existingDocumentWithID:] not returning nil for no…

    …nexistent doc
    
    (Reported in post to mailing list)
    authored March 27, 2014
  2. Jens Alfke

    Added CBLIncrementalStore source files to Extras

    The “Extras” folder generated by the iOS build target now includes CBLIncrementalStore.{h,m}.
    Also, the README in that folder has been updated to describe it.
    authored March 27, 2014
  3. Jens Alfke

    Back to $ARCHS_STANDARD for iOS builds

    Removes the “Release-64bit” build configuration.
    With Xcode 5.1+, the regular iOS frameworks will include 64-bit.
    With earlier Xcodes only 32-bit.
    NOTE: This commit breaks the automated Jenkins build.
    The build script will need to remove the steps to build with Release-64bit,
    and to copy extra 64-bit frameworks.
    authored March 27, 2014

Mar 25, 2014

  1. Jens Alfke

    Declare public C functions as `extern “C”` when compiled in C++ mode

    Otherwise Obj-C++ apps get link errors using functions like CBLTextKey.
    Fixes #291
    authored March 25, 2014

Mar 24, 2014

  1. Jens Alfke

    Tweaked logging in CBLReplication.m some more.

    authored March 24, 2014
  2. Jens Alfke

    Finally fixed the multipart JSON-decoding bugs

    Fixes #288
    Fixes #289
    authored March 24, 2014
  3. Jens Alfke

    Tweaked some logging in CBLReplication.m.

    authored March 24, 2014
  4. Jens Alfke

    Fixed another assertion-failure when a _bulk_get download can’t parse…

    … a JSON doc
    
    Found a piece that was missing from the prior fix 24f6381. The same assertion failure was occurring,
    under slightly different circumstances.
    Fixes #288 (speculatively). Also see #271, #232.
    authored March 24, 2014
  5. Jens Alfke

    Fix for when transformer block rejects a pulled doc

    Needed to mark the doc's sequence as no longer pending. Otherwise the
    checkpoint doesn't advance, and the puller will keep trying to pull that
    doc on subsequent runs.
    authored March 22, 2014

Mar 20, 2014

  1. Jens Alfke

    Test class names should start with “CBL”

    They’re baked into debug builds of the framework, and I just ran into a name conflict
    when trying to build a developer’s app with a debug CBL, because he had a class
    called “TestModel” too.
    authored March 20, 2014
  2. Jens Alfke

    Thread-safety fix in CBLWebSocketChangeTracker

    I haven’t seen this cause problems, but fixed it just in case.
    authored March 20, 2014
  3. Jens Alfke

    Replication encoding/encryption improvements

    * Transformation block can now return nil to skip the document (useful on pull if it’s invalid)
    * Assign correct revpos to an attachment created by a transformer during a push
    authored March 20, 2014
  4. Jens Alfke

    CBLModel improvements (-willSave:, empty _attachments)

    * Made -willSave: public (it was always meant to be but I forgot to add it to the header)
    * Fixed some situations where a model could end up with an _attachments dictionary with zero elements
    * -propertiesToSave wasn’t including _id prop if doc was new; caused trouble for signed documents.
    authored March 20, 2014

Mar 19, 2014

  1. Jens Alfke

    Puller was getting confused receiving JSON-only docs (no attachments)…

    … with gzip response encoding
    
    It was looking at the “Content-Encoding: gzip” header and thinking that this was a JSON _part_ that
    was encoded, so it tried to decode it itself. Whereas the top-level gzip decoding is done already by
    NSURLConnection. So it was trying to gzip-decode plaintext, oops.
    Fixes #281
    authored March 19, 2014
  2. Jens Alfke

    Oops, removed some leftover debugging code from the last commit

    authored March 18, 2014

Mar 18, 2014

  1. Jens Alfke

    WebSocket changes feed was sending ‘since’ to SG as string, not number

    This unfortunately caused the changes feed to start from the beginning every time.
    authored March 18, 2014
  2. Jens Alfke

    Optimizations for view indexing and querying

    These shave a significant amount of time off of Ashvinder’s Test 12.
    authored March 17, 2014

Mar 16, 2014

  1. Jens Alfke

    Fixed incorrect values from CBLUnsavedRevision and CBLAttachment, for…

    … unsaved attachments
    
    CBLUnsavedRevision:
    -attachmentMetadataFor: and -attachmentNamed: would return nil for an unsaved attachment.
    -attachments wouldn’t return any unsaved attachments.
    
    CBLAttachment:
    .length would return 0 for an unsaved attachment.
    authored March 16, 2014

Mar 15, 2014

  1. Jens Alfke

    Removed all deprecated API methods.

    authored March 15, 2014
  2. Jens Alfke

    Replication encoding/encryption API now supports attachments.

    During a pull, the _attachments objects for large attachments will have
    a temporary "file" property that points to the local file path where
    the attachment data can be found. The transformation block can read this
    file, but must not modify it.
    authored March 14, 2014

Mar 13, 2014

  1. Jens Alfke

    Fixed WebSockets bug with zero-length messages

    The WebSocket class broke when it received a zero-length message, which
    is what the _changes feed sends when it's caught up.
    Fixes #272
    authored March 13, 2014

Mar 12, 2014

  1. Jens Alfke

    Fixed a couple of minor warnings reported by Xcode 5.1

    authored March 12, 2014
  2. Jens Alfke

    Don’t set replication progress to 0/0 when stopping

    Instead the progress should remain the way it was.
    Also, improved Traun’s great new test for verifying that progress counts make sense.
    Fixes #274
    authored March 11, 2014
Something went wrong with that request. Please try again.