Accidental

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>> Well, I know 'pqm-submit' takes the extra step to check if you have
>> uncommitted changes, and by default aborts. It wouldn't be hard to have 'bzr
>> send' do a 'is working tree clean' check first. It *might* make it a little
>> slower, is the only real reason not to.
> 
> The other reason is that it prevents people from *deliberately* sending
> when they have changes in their wt.  I'd rather solve it by just
> refusing to send when the user has implicitly specified an empty merge
> directive.  I'd even let them specify an empty merge directive, if they
> specified it explicitly.
> 
> I don't see a reason to be so strict, especially when it slows down the
> process.
> 
> Aaron

I've actually been saved by pqm-submit a couple times, and I would guess other
people have as well.

I would say it is probably the minority case that you have a branch where you
are continuing to work and have uncommitted changes that you don't want to send.

You could always send anyway with a flag (such as --force). But at least it
would be a helpful "are you sure, you have some stuff you forgot."

I think we need the empty merge directive check anyway, because of all the
times people accidentally use the source branch as the submit branch (bzr send
. pretty much never does what you want, but it is tempting.)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIe67MJdeBCYSNAAMRAgFtAKCJ4rBsiPEADbi2dDlzFJzucQhc7QCdEGqz
KHKzlnKaVfJ0Qw76CiGtbEM=
=UfkN
-----END PGP SIGNATURE-----