view release on metacpan or  search on metacpan

Changes  view on Meta::CPAN

      local path is taken off the native transport, because libgit2's local
      transport pushes to the remote's fetch URL rather than the URL it
      connected to -- writing `refs/karr/*` into the wrong repository and
      calling it success; the CLI fallback lands at the push URL, and the
      detection is narrow enough to leave every real transport to libgit2
      (k208).

    - `karr sync` reports what a fetch moved correctly against libgit2 1.9,
      by requiring `Git::Native 0.006`. 0.006 takes `git_fetch_options.prune`
      from `Git::Libgit2::fetch_options_prune_offset` rather than a constant
      derived from libgit2 1.5; against 1.9 -- where the embedded callbacks
      struct is 128 bytes, not 120 -- that constant put the prune value eight
      bytes early, on the `update_refs` callback pointer 1.9 prefers over
      `update_tips`. `App::karr::Git::_fetch_refspecs` reads the fetch
      result's `updated` list to decide what a sync moved, so a fetch that
      did report ref updates was the case that misread it.

    - Parallel writes and fresh clones stopped losing races (k172, k173,
      k181, k182, k183). Eight `karr create` at once handed id 8 to two
      cards, because a pull adopted a remote counter behind the local one;
      the counter is now merged, taken only when ahead and compare-and-swapped

lib/App/karr/Git.pm  view on Meta::CPAN

# refspec can never apply one. push has always been forced; pull was not, and
# libgit2 declines a non-ff fetch update without raising an error, so pull
# returned success while leaving the ref stale -- and the next push then
# force-wrote that stale ref over the other agent's work (#40). Both
# directions are forced now; see _fetch_refspec for the pull side.
#
# The semantics this settles on are last-writer-wins, which is what the
# parentless-commit design already implied everywhere else. Doing better
# would need compare-and-swap on the ref (git_reference_create_matching,
# unbound in Git::Libgit2 -- see ticket #81) plus per-ref rejection reporting
# from libgit2's update_tips/push_update_reference callbacks (not installed
# by Git::Native -- ticket #80). Neither is reachable from karr today.
use constant BOARD_REFSPEC => '+refs/karr/*:refs/karr/*';

use constant BOARD_ROOT => 'refs/karr/';

# The board's identity (#95): stamped once at board birth, compared on every
# pull before any reconciliation -- see _check_board_identity. Declared with
# the other namespace constants because use constant is only visible from its
# textual point on, and _check_board_identity needs it.
use constant BOARD_ID_REF => 'refs/karr/meta/board-id';

lib/App/karr/Role/TaskMutation.pm  view on Meta::CPAN

# same string; this role needs it directly because it reads the OID and the
# content together (App::karr::Git::read_ref_with_oid), which is the pair a
# compare-and-swap has to guard against, and no BoardStore method hands both
# back.
sub _task_data_ref {
    my ($self, $id) = @_;
    return "refs/karr/tasks/$id/data";
}

# The one way a callback can say "there is nothing here to write". A
# package-lexical scalar ref, compared by address, because the callbacks on
# this path return whatever their last statement happened to evaluate to -- a
# title, a status, the result of a clearer -- and no value a caller can
# construct by accident is this one.
#
# The address is read with Scalar::Util::refaddr, which is how identity is
# asked in this distribution (App::karr::SyncGuard keys its registry of armed
# guards the same way). It also settles the overloading question instead of
# stepping around it: refaddr returns the address itself and never consults an
# overloaded `==`, so a blessed return value cannot be asked a comparison it
# would answer with an exception.