Switch branches/tags
Commits on Mar 28, 2012
  1. Remove assertion when database compaction finishes

    fdmanana authored and janl committed Mar 28, 2012
    Because the task status is updated only if the time elapsed
    since the last task status update is greater than some treshold,
    the value of the changes_done property might be smaller than
    Closes COUCHDB-1451
Commits on Mar 24, 2012
  1. travis: make distcheck

    rnewson committed Mar 24, 2012
Commits on Mar 23, 2012
  1. Configure for travis.

    rnewson committed Mar 22, 2012
  2. log 500 errors before return them to the client

    tilgovi committed Mar 23, 2012
    This ensures that the log will contain any nasty errors we haven't
    caught elsewhere. In particular, related to the previous patch, it
    ensures that couch_file errors otherwise only logged at debug will
    log at error unless caught elsewhere.
  3. lower couch_file error logging on open to DEBUG

    tilgovi committed Mar 23, 2012
    The reason is that, while possibly informative when something does
    need debugging, in the common case where the caller means to create
    the file if it doesn't exist this may not be a genuine error.
    Perhaps better would be if there were a way to actually request that
    the file be created without requesting it be overwritten.
Commits on Mar 21, 2012
  1. maintain the file_exists idiom

    tilgovi committed Mar 21, 2012
    POSIX {error, eexist} from the file module needs to be kept from
    bubbling all the way to couch_httpd or it will result in a 500 error
    when a 412 is what we want.
    This change confines all the "database exists already" conversion of
    the {error, eexist} that now bubble out of couch_file contained within
  2. no really, I was mostly right

    tilgovi committed Mar 21, 2012
    The errors that come back from the file module are always of the form
    {error, Reason}, and Reason is the atom that should be passed to
    In one case, couch_file was returning the atom file_exists rather than
    comforming to the file module's errors I was expecting.
    Returning {error, eexist} is cleaner.
  3. Revert "Receive the correct error message to proceed upon ignoring a …

    tilgovi committed Mar 21, 2012
    …file-open error"
    Most of the errors here will be of the form {error, Reason} because
    they come from the file module. Instead of minting our own file_exists
    atom lets use {error, eexist}.
    This reverts commit 3e5ad08.
  4. Receive the correct error message to proceed upon ignoring a file-ope…

    Jason Smith (air)
    Jason Smith (air) committed Mar 21, 2012
    …n error
Commits on Mar 20, 2012
Commits on Mar 19, 2012
  1. Fix error stack traces

    davisp authored and rnewson committed Nov 22, 2011
    If you get the stack in an exception handler after calling a function it
    gets changed from where the exception was thrown. Fixed simply by
    grabbing the stack before making logging calls.
  2. don't delete view groups that fail to open

    tilgovi committed Mar 18, 2012
    ...but log a message instead. Fixes COUCHDB-1445.
  3. clear ETS table entry if view group fails to start

    tilgovi committed Mar 19, 2012
    Without this, the next process which tries to open the view group
    will add itself to the wait list even though the monitor has already
    fired. When that happens, the caller hangs forever.
  4. remove unnecessary eaccess special casing

    tilgovi committed Mar 19, 2012
    It's better to let these errors bubble and/or not give them special
    treatment when file:format_error/1 can do a better job of describing
    the failure.
Commits on Mar 18, 2012
  1. COUCHDB-1426 Ensure use of specified SpiderMonkey

    davisp committed Mar 14, 2012
    There were a few bugs in configuration settings when users specified
    --with-js-include and --with-js-lib for custom SpiderMonkey
    installations when a SpiderMonkey library is installed globally. This
    patch addresses this by reordering configuration checks to use the
    specified variables without adding information from pkg-config checks or
    from default settings.
    There is also a new setting --with-js-lib-name which is required due to
    SpiderMonkey having so many names for its libraries. If configuration or
    building fails the most likely scenario is that it picked up an older
    installed SpiderMonkey in the library search path. To fix you should
    only need to add --with-js-lib-name with the value that should be passed
    to the linker. Ie, -lfoo would be set as "--with-js-lib-name=foo".
Commits on Mar 13, 2012
  1. [docs] bring README, THANKS, AUTHORS up to date for 1.2.0

    dch committed Mar 13, 2012
    - correct Firefox test suite usage notes
    - add jhs@
  2. Fix view etag calculation regression introduced by fb7225

    janl committed Mar 13, 2012
    The optimisation relied on creating batches of documents to be
    sent to the view server. The implementation created the batch
    in a foldl function instead of a foldr function causing the
    wrong update sequence to be persisted because the write assumes
    the last item in the list to have the largest sequence.
    Thanks to Filipe for guiding the research.
Commits on Mar 12, 2012
  1. Fix very slow test test/etap/220-compaction-daemon.t

    janl committed Mar 12, 2012
    After the recent change in the view updater that ensures
    the view updater's writer process collects larger batches
    of map values to write into the index, the test's desired
    fragmentation of 70% takes much more time to happen.
    To fix this, simply decreased the target fragmentation.
    (patch by Filipe)
  2. update NEWS and CHANGES

    janl committed Mar 12, 2012
Commits on Mar 8, 2012
  1. Add 100ms sleep to etap:end_tests() to prevent hanging

    bitdiddle committed Mar 8, 2012
    It's not clear if the issue is etap, escript, prove, or couchdb. This
    needs further investigation.
  2. Whitespace

    bitdiddle committed Mar 8, 2012
  3. [docs] compression is enabled by default and bumps on-disk version

    dch committed Mar 8, 2012
    - updated NEWS and CHANGES
Commits on Mar 5, 2012
  1. More efficient view updates

    fdmanana committed Mar 5, 2012
    In several scenarios the writer process of the view
    updater was collecting very small batches of map
    values, which is not optimal regarding efficiency.
    This change ensures the batches are larger by queing
    batches of map values into the writer's queue instead
    of queing map values one by one.
    This queing behaviour was changed by COUCHDB-1186 and
    its non-optimal efficiency for certain documents/map
    functions wasn't noticed before.
Commits on Mar 3, 2012
  1. update NEWS & CHANGES

    janl committed Mar 3, 2012
Commits on Mar 2, 2012
  1. Merge branch 'R15B-driver' into 1.2.x

    janl committed Mar 2, 2012
    * R15B-driver:
      Make icu_driver compatible with interface changes in R15B.
  2. Make icu_driver compatible with interface changes in R15B.

    janl committed Feb 27, 2012
    Erlang R15B redefines* the signature for the driver control callback to
    use its own special ssize_t type. This just adds a conditional based on
    the definition of ERL_DRV_EXTENDED_MAJOR_VERSION to pick the right type.
Commits on Feb 22, 2012
  1. Improving README

    nslater committed Feb 22, 2012