[MERGE] Make mail-to address in ``bzr send`` optional for interactive mail clients

John Arbash Meinel john at arbash-meinel.com
Mon Nov 26 20:50:38 GMT 2007


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

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>> Lukáa Lalinský wrote:
>>> Hi,
>>> if bzr send is used with an interactive mail client, it's easier to fill
>>> in the target address in the client rather than on the command line
>>> (address book, auto-complete, etc.). I think it makes sense to make it
>>> optional in such cases. For the 'editor' client, which sends mail
>>> directly, it's still required.
> 
>> I believe the default for "bzr send" is to write to stdout. So it may be weird
>> to have plain "bzr send" sometimes write to stdout and sometimes spawn a mail
>> editor depending on the config settings.
> 
> No, the default is to invoke a mail client.  Which may be a
> fully-interactive client like t-bird, or may be the "editor" client.
> 
>> This is certainly an argument in favor of requiring -o-. AIUI, we retained "bzr
>> send > foo.patch" behavior because it was the old bundle behavior. But then
>> didn't we split bundle out into its own command again?
> 
> I retained "bundle" because I didn't think it made sense to make it an
> alias of "send" if "send" had different behavior.  "bundle" is
> implemented as a subset of "send".
> 
>> So maybe we should switch back the "must supply -o or --mail-to" behavior
>> (where it can be obtained from config settings), and make it obvious that it is
>> a command line api break.
> 
> I don't really understand your reasoning.  Is it based on the
> misconception that the "send" command's behavior varies?  The only thing
> that varies is which client is selected.
> 

Probably remembering old effects from my experience when "bzr bundle" was an
alias for "bzr send".

>> I think Aaron's comment is that you can do something like (this is my config):
> 
>> [/home/jameinel/dev/bzr]
>> submit_branch=/home/jameinel/dev/bzr/bzr.dev
>> submit_to = Bazaar Mailing List <bazaar at lists.canonical.com>
> 
>> [/home/jameinel/dev/bzr/bzr.dev]
>> public_location=http://bazaar-vcs.org/bzr/bzr.dev
> 
> Yes, but actually people have suggested that the submit_to value should
> be associated with [/home/jameinel/dev/bzr/bzr.dev] rather than the
> source branch.
> 
> Aaron

I think it could pick it up from there, but if a branch has its own submit_to
it should be preferred.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHSzGeJdeBCYSNAAMRAvdmAJ0cPuIwiHrUuMKtDXvjUMqx10Y1kgCffyaq
V5Dk3rSarm3HyMhtvO3bAFw=
=k36m
-----END PGP SIGNATURE-----



More information about the bazaar mailing list