"C. Michael Pilato" <cmpilato@collab.net> writes:
> I used to be in favor of this, but now I'm not so sure. The idea of a shell
> mode for 'svn' that seems most intuitive to me is simply an interactive way
> to do all the other things that 'svn' can do today with multiple subcommand
> invocations. But MUCC is strictly about doing a single particular task that
> you *can't* do with a string of 'svn' subcommands today. A shell mode for
> 'svn' probably needs to be a little more full-featured, able to examine
> working copies and repositories alike -- MUCC's use-case doesn't care about
> working copies. Also, I need MUCC to maintain a non-interactive mode.
>
> If all these things can be satisfied by introducing 'svn shell', that's cool
> -- I don't think I'd fight it. But I guess I'm thinking "baby steps" here.
> contrib/client-side/mucc.c --> subversion/svn-mucc/main.c is a
> super-trivial change. And in anticipation of this move, I even recently
> rewrote the mucc test script to sorta-kinda use our Python test framework.
>
> If folks are cool with introducing svn-mucc, I'm volunteering for the task.
The point of a shell mode is that it all happens in one txn (which is
exactly the point of mucc, AIUI). You do a bunch of things, and then
you "commit". Any shell mode is also a non-interactive scripting mode
too -- just feed the commands on the shell's stdin.
Given this new understanding of the proposed svn shell mode, do you
still prefer mucc?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Mar 16 02:51:29 2007