Pseudorefs store transient data in in the repository. Examples are HEAD,
CHERRY_PICK_HEAD, etc.
These refs have always been read through the ref backends, but they were written
in a one-off routine that wrote an object ID or symref directly into
.git/<pseudo_ref_name>.
This causes problems when introducing a new ref storage backend. To remedy this,
extend the ref backend implementation with a write_pseudoref_fn and
update_pseudoref_fn.
Signed-off-by: Han-Wen Nienhuys <hanwen@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
d05c8e45f1