Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on Mar 4, 2013
  1. Benoit Chesneau

    report a more accurate progress.

    benoitc authored
    Instead of reporting only when the sequence is commited report also when a
    sequence is processed. It helps the user to figure what's going on.
    Note: this is only useful on normal replications, since on continuous
    replications the number of sequences to process will change over the time.
  2. Benoit Chesneau

    support erlang R16

    benoitc authored
Commits on Feb 10, 2013
  1. Benoit Chesneau
Commits on Feb 9, 2013
  1. Jan Lehnardt Benoit Chesneau

    refactor replicator_db_security to be more resilient

    janl authored benoitc committed
  2. Robert Newson Benoit Chesneau

    support JSONP in externals

    rnewson authored benoitc committed
  3. Benoit Chesneau

    Unfix view compaction progress reports

    benoitc authored
    This reverts the compaction task status progress back to what exists on
    1.2.x. The issue is that we use the number of documents in the database
    instead of the number of docids in the union of all views (the number of
    rows in the view's id btree). In this particular case the desire to have
    seamless view upgrades outweighs fixing the relatively minor UI bug.
    backport from couchdb 1.3.x branch
  4. Benoit Chesneau

    COUCHDB-1654: Transparently update view signatures from <= 1.2.x.

    benoitc authored
    Updates 1.2.x or earlier view files to 1.3.x or later view files
    transparently, the first time the 1.2.x view file is opened by
    1.3.x or later.
    Here's how it works:
    Before opening a view index,
    If no matching index file is found in the new location:
     calculate the <= 1.2.x view signature
     if a file with that signature lives in the old location
       copy it to the new location with the new signature in the name.
    Then proceed to open the view index as usual.
    After opening, read its header.
    If the header matches the <= 1.2.x style #index_header record:
      upgrade the header to the new #mrheader record
    The next time the view is used, the new header is used.
    If we crash after the rename, but before the header upgrade,
      the header upgrade is done on the next view opening.
    If we crash between upgrading to the new header and writing
      that header to disk, we start with the old header again,
      do the upgrade and write to disk.
    backport from Couchdb 1.3.x branch
  5. Benoit Chesneau

    COUCHDB-1667: Revert index_dir to view_index_dir config option.

    benoitc authored
    Patch by Alexander Shorin. (backport from couchdb 1.3.x branch)
  6. Jan Lehnardt Benoit Chesneau

    Do not log errors on creating new view index.

    janl authored benoitc committed
    While view index file may be really missed, it's ok for new views
    and this error will be self-healed on next couch_file:open call
    so there is no reason to worry user about.
    Patch by Alexander Shorin.
  7. Jan Lehnardt Benoit Chesneau

    Avoid badmatch when replicator dbnames have leading slashes.

    janl authored benoitc committed
    Closes COUCHDB-317
    Patch by Joan Touzet.
Commits on Jan 26, 2013
  1. Benoit Chesneau

    prevent crash

    benoitc authored
Commits on Jan 8, 2013
  1. Benoit Chesneau

    fix view seq rettrieving

    benoitc authored
  2. Robert Newson Benoit Chesneau

    fix pbkdf2 spec

    rnewson authored benoitc committed
  3. Robert Newson Benoit Chesneau

    Delete view files on database deletion

    rnewson authored benoitc committed
    couch_file:nuke_dir attempts to recurse into subdirectories and delete
    the contents, it does so when couch_file:delete returns {error,
    eperm}. Unfortunately, the subdirectory has been renamed to a random
    uuid before returning, and so the recursive call gets {error, enoent}
    from list:dir(Path) where Path is the original (non-existent)
    subdirectory path.
    View files are not currently deleted when a database is because of
    this (this happened since 1.2.0, no release is broken) because the
    view engine rewrite added a further directory level called 'mrview'.
    This patch modifies nuke_dir to depth-first, solving the issue.
  4. Robert Newson Benoit Chesneau

    Ignore multipart epilogue

    rnewson authored benoitc committed
  5. Robert Newson Benoit Chesneau

    Include a comment before jsonp output

    rnewson authored benoitc committed
  6. Robert Newson Benoit Chesneau

    Improve script url validation

    rnewson authored benoitc committed
Commits on Dec 18, 2012
  1. Benoit Chesneau

    fix path to get last sequence

    benoitc authored
Commits on Dec 8, 2012
  1. Benoit Chesneau
  2. Benoit Chesneau
  3. Benoit Chesneau
  4. Benoit Chesneau

    fix badarg error

    benoitc authored
  5. Benoit Chesneau

    make sure that the vhost server is started.

    benoitc authored
    wait for the vhost server start before to call it.
  6. Benoit Chesneau

    fix error format

    benoitc authored
  7. Adam Kocoloski Benoit Chesneau

    Close view files cleanly on DB create/delete

    kocolosk authored benoitc committed
    The shutdown_sync was actually interfering with a clean shutdown
    initiated by the index processes themselves when they received a 'DOWN'
    message from their monitoring of the DB process.  The explicit 'delete'
    cast message in this patch is very probably redundant, but it makes the
    intent clear.  The important bit is to let the index process terminate
    *and* get confirmation of the termination before attempting to nuke the
  8. Benoit Chesneau

    missing validate

    benoitc authored
Commits on Dec 5, 2012
  1. Benoit Chesneau

    don't use cowboy.

    benoitc authored
    cowboy was introducing some issue and isn't enough reliable for our
    purpose so for now only use ranch to improve the accepting loop and
    reduce memory and fd leaks.
Commits on Dec 4, 2012
  1. Benoit Chesneau

    useless var

    benoitc authored
  2. Caleb Case Benoit Chesneau

    COUCHDB-430,514,764 Fix list HTTP header handling.

    calebcase authored benoitc committed
    Currently calls to getRow() cause the HTTP headers to be sent immediately back
    to the client. This happens even if an error is thrown after the getRow(), but
    before any send(...) or start(...). Worse, if a list throws an exception an
    extra, invalid header is sent to the client (resulting in various bad
    Erlang list handling will now wait until data has been sent BEFORE sending the
    HTTP headers to the client. If an error is reported it will result in an HTTP
    error code as expected. This does not change the behavior of errors thrown
    AFTER data has been sent: They will still result in an HTTP 200 even if an
    error is reported.
    The line protocol between Erlang and os processes has been extended to support
    an optional Header field on "chunks" and "end". The javascript list handling
    has been updated to use this if a new header is set via start(...). This makes
    it possible to begin processing with getRow(), but later reset the headers via
    start(...). Again, if data has been sent(...) the new headers will NOT take
Commits on Nov 28, 2012
  1. Benoit Chesneau
  2. Benoit Chesneau

    load config from a list of config files set in the app env. fix #18

    benoitc authored
    Instead of having a fixed list of files we are now loading them from a
    list of config files. Defaults are now:
        <coderoot>/etc/{couch.ini, couch_httpd.ini, local.ini}
Commits on Nov 27, 2012
  1. Robert Newson Benoit Chesneau

    Return X-Couch-Id from show functions if doc is created

    rnewson authored benoitc committed
  2. Benoit Chesneau

    Revert "Document Id and Rev in response headers"

    benoitc authored
    This reverts commit 5fd8380.
  3. Robert Newson Benoit Chesneau

    test for 1608

    rnewson authored benoitc committed
  4. Robert Newson Benoit Chesneau

    Support auth cookies with : characters

    rnewson authored benoitc committed
    The parts of a couchdb authentication cookie are separated by
    colons. One of these parts can contain colons and, more rarely, runs
    of colons. The string:tokens function silently drops any empty token,
    thus giving a spurious failure for valid input. The fix changes this
    mechanism to one that losslessly decodes this part.
Something went wrong with that request. Please try again.