[RFC] I want to disable submit_branch on my computer for all branches. How can I do that?

[vila]

>>>>> Eric Siegerman <lists08-bzr at davor.org> writes:

    > I've also suffered from Alexander's submit-branch frustration.
    > I've never tried to analyze it though, just (with annoyance)
    > deleted the config setting whenever it happens to be wrong,
    > and hoped it would *stay* unset.

It will... until you merge again :-} 

Or may be not I'm not sure if setting it to '' is the same as not
setting it all but I doubt the code handles them differently (may be it
should).

    > On Thu, 2011-05-12 at 14:31 +0200, vila wrote:
>>>>> Alexander Belchenko <bialix at ukr.net> writes:
    >> I think the truth is that [the submit branch] *defaults* to
    > parent_location if it's not
    >> set.

    > Correct.

    >> > The problem here is that submit_branch only set by merge,
    >> 
    >> That's indeed the source of the problems.

    > Perhaps the solution is to have submit_branch *not* be set by
    > merge, i.e. make "bzr merge --no-remember" the default, even if
    > submit_branch is not currently set.

Will fit my workflow I think.

The more I think about config options and --remember the more I think
--remember should be re-designed or at least configurable by the user.

    > Orthogonal to that is another possible change:  don't default
    > submit_branch to parent_location; if the former is unset, it
    > doesn't point to anything, and attempts to reference it fail.

-0.5, I'd prefer to fix its usage, even if we manage to get it wrong
only for the edge cases, it's better than failing most of the time.

    > Instead, have "bzr branch" *explicitly* set submit_branch to the
    > same value as parent_location.  That gives the same effective
    > results as now in many cases,

Well, you're trading one guess for another, can't we find ways to
enhance our guess instead ?

It's harder to make good guesses than add options but when we make good
guesses the overall experience is better.

        Vincent