Commits on Nov 8, 2012
  1. Release 0.08204

    Getty committed Nov 8, 2012
Commits on Nov 5, 2012
Commits on Nov 4, 2012
Commits on Nov 3, 2012
  1. Fix API mismatch between new_related() and new_result() (RT#78336)

    Almost 7 years ago a refactor in fea3d04 fed an (undocumented) $attrs
    parameter from new_related() to new_result(), while new_result() never
    expected (and ignored) said parameter. Since this is an undocumented
    feature, which nobody complained about for all this time - document
    extensively and kill it with fire.
    ribasushi committed Nov 3, 2012
  2. Fix unique constraint violations in blanket movement (RT#7…

    This essentially reverts 5e6fde3 which was applied without much thinking.
    Added a test to ensure this doesn't re-break again.
    ribasushi committed Nov 3, 2012
  3. Properly signify internal cursor methods, trash their accessors as well

    (the accessors were not used throughout anyway)
    ribasushi committed Sep 14, 2012
  4. fix Cursor SYNOPSIS

    rkitover committed with ribasushi Aug 26, 2012
  5. Let SQLMaker rs_attr 'for' support string literals

    SQLMaker previously only allowed hardcoded values with the 'for' attr,
    overriding in storage specific subclasses. Rather than attempt to provide an
    exhaustive list of possible options, the base class can now take \$scalaras
    an override that is embedded directly in the returned $sql
    mattp- committed with ribasushi Oct 19, 2012
  6. Fix invalid pod example syntax

    Felix Antonius Wilhelm Ostmann committed with ribasushi Oct 20, 2012
Commits on Oct 20, 2012
  1. release 0.08203_01

    ribasushi committed Oct 20, 2012
Commits on Oct 18, 2012
  1. release 0.08203

    frioux committed Oct 18, 2012
Commits on Oct 15, 2012
  1. release 0.08202_01

    frioux committed Oct 15, 2012
Commits on Oct 12, 2012
  1. Really fix the DBD::SQLite ping() issue

    Apparently DBD::SQLite is notoriously bad at synchronizing its internal
    transaction state with {AutoCommit}
    There is a function
    but DBD::SQLite does not expose it (nor does it seem to properly use it)
    Furthermore the detection is rather broken as described in RT#80087.
    Bend over backwards to attempt to preserve as much sanity as possible.
    While at it issue a non-trappable warning so that folks fix the offending
    codepath (which arguably is still broken)
    It is possible to have a proper "connection", and have "ping" return
    false anyway (e.g. corrupted file). In such cases DBD::SQLite still
    keeps the actual file handle open. We don't really want this to happen,
    so force-close the handle via DBI itself (this solves a bunch of Win32
    test failures)
    ribasushi committed Oct 9, 2012
Commits on Oct 6, 2012
Commits on Oct 5, 2012
  1. release 0.08201

    frioux committed Oct 5, 2012
  2. Replace inadequate $dbh->ping SQLite implementation (RT#78420)

    When SQLite attempts to connect to a file that is not a database,
    it nevertheless maintains a true $dbh->{Active} and $dbh->ping.
    Replace with a schema listing SELECT, and fix test erroneously
    assuming it can portably do chmod 000
    ribasushi committed with frioux Aug 30, 2012
Commits on Sep 1, 2012
  1. Revert "Fix t/storage/reconnect.t when run as root"

    The test change just papered over the problem, a proper fix is pending
    on the for_current/riba/fix_sqlite_ping_RT78420 branch.
    This reverts commit 64819b0.
    ilmari committed Sep 1, 2012
  2. Fix t/storage/reconnect.t when run as root

    As root, the chmod 000 doesn't prevent SQLite form opening the fake db
    file, so we need to disconnect to make it reopen the restored real one.
    ilmari committed Sep 1, 2012