Re: [Gnu-arch-users] Re: [BUG] feature plan -- merge request headers

[Tom Lord]

    > From: Stefan Monnier <address@hidden>

    > > Now someone can fix a trivial bug in arch with:

    > >         % address@hidden

    > >         % tla get $LORD/dists--devo--1.0 wd
    > >         % cd wd
    > >         % tla buildcfg emf.net/devo
    > >         % cd src/tla
    > >         % tla fork bug123
    > >         [hack hack hack]
    > >         [test test]
    > >         [hack]
    > >         [test]
    > >         [hack]
    > >         % tla commit --announce

    > But doesn't the `fork' step require write access to the
    > address@hidden archive?  This is more a question about `fork' I guess,
    > but mypoint is that it's important for `fork' to be able to make a fork in
    > some other archive, which also means that it can't so easily benefit from
    > the "downstream branch" feature (unless that feature is extended so it can
    > relate two branches in two different archives).


No, `fork' in that case would create the new branch in the user's
default archive.

Or something like that, anyway --- it is a bit awkward that so many
other commands want to default to using the archive of the project
tree in `pwd` but fork is one that shouldn't.  Or should it?

-t