[PATCH 00 of 18] Current advance on the states/phase topic

pierre-yves.david at logilab.fr pierre-yves.david at logilab.fr
Mon Oct 10 07:27:56 CDT 2011


Alain Leufroy and I are currently working on a states/phases implementation for
mercurial-core.

Here is a first series that display the current state (phase ?) of our work for
preview and feedback. This series is not meant to be applied.

If Matt Mackall feels this feature can make it into 2.0 we shall we able to
provide resource to make it happen.

Notes
-------

* States have been renamed to phases to avoid confusion with Status

* We use phase names as few as possible until proper name are picked. This is to
  be changed in the final series for clarity. See the link bellow for details
  http://mercurial.selenic.com/wiki/StatesPlan#Naming

  the revset filtering on phase suffer the most from this lack of name.


Future works
------------

* implement phases boundary exchange and publish=False server: Please see my
  previous email on the list about it.

* Adding a phase: <phasename> when phase != 0. It's pretty trivial but I prefer to wait for phasename to be choosed before altering all test with a new log output.

* Behaviour of current extension will change:

    - rebase will refuse to work on public changeset without --keep.
    - mq will refuse to qimport public changeset.
    - mq managed changeset will be secret.

  The code is not written yet but it does not seems too complicated.

* The code to move the boundary backward exist in the experimental extension: 
  But have not been ported yet.

   https://bitbucket.org/marmoute/mutable-history/changeset/d5170cc7881c




-- 
Pierre-Yves

PS: Alain and I are using the experimental implementation of evolution to
    collaborate on this.


More information about the Mercurial-devel mailing list