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

[vila]

>>>>> Martin Pool <mbp at canonical.com> writes:

    >> > Let's see:
    >> >
    >> > no exiting setting:
    >> > * default: current setting is stored,
    >> > * --remember: current setting is stored,
    >> > * --no-remember: current setting is not stored
    >> >
    >> > existing setting:
    >> > * default: current setting is *not* stored,
    >> > * --remember: current setting is stored
    >> > * --no-remember: current setting is not stored
    >> >
    >> > Yes, I cheated, I used None, True, False, thanks python :)
    >> >
    >> > Does that work for you ?
    >> 
    >> I don't like it, because it breaks "merge --remember --no-remember" when
    >> there's no existing setting.

    > I'd actually go a bit further than Vincent and say that perhaps remembering
    > things depending on whether there is already a default is complicating the
    > user model more than is worth while for the convenience.

    > How about instead:

    > Locations are never remembered by default. (Perhaps they should be set while
    > calving branching.)

Well, *I like* the way bzr helps me not repeating myself ! merge is the
only exception but adding a merge_branch would address both my concerns
and Alexander's ones I think (as long as the user retains control about
when it's set).

    > With --remember (can be canceled with --no), the location is stored as the
    > default.

    > With --no-default-location (also can be canceled), the command fails if no
    > explicit location is given.

Hmm, adding another option sounds more complicated for the user IMHO.

I think the issue here is that the *default* behavior when no previous
setting exists is to remember, *whether or not* --remember or
--no-remember is specified.

i.e. cmd_merge has remember=False as a parameter but ignores it if there
is no previous setting. I'm fine with that *iff* we change it to
remember=None.

I argue that the default behavior should obey this.

Aaron argue that (without previous setting):

 bzr merge --remember --no-remember 

should still remember the setting.

But this is precisely what Alexander (and others including me) is
complaining about (and I don't understand the use case it's addressing,
if the user is explicit about his desire (--no-remember specified last),
we should respect it).

So my vote will be to:

- add a merge_branch (or bzr.merge.from) setting (still defaulting to
  parent or should bzr complains there ?)

- change the default behavior to respect user input.

  Vincent