[PLUGIN] pqm-submit

[John Arbash Meinel]

Robert Collins wrote:
> I've had a play with it... it doesn't fit the way I'm working, and I
> think it is indicative of a more general issue...
> 
> 'parent' for me is http://bazaar-vcs.org/bzr/bzr.dev (using the new
> combined web site url, which I'm going to stash into pqms config later
> today).
> 
> Then I do 'bzr pull --overwrite' to sync with the shared branch
> regularly - I'd use a bound branch if we could teach it to commit via
> pqm ;).
> 
> so the public location for me
> 'http://people.ubuntu.com/~robertc/baz2.0/integration'
> which isn't trivially guessable from my push location:
> hostname:public_html/baz2.0/integration which is soon to be
> sftp://hostname/~/public...
> 
> I think pqm-submit should use the parent as the as default target (I may
> have misread the code and it does this now.. correct me if I'm wrong).
> 
> And we can/should have some rule (i.e. a sed rule) to convert the branch
> push location to the public location.
> 
> Perhaps a prefix-replacement is enough:
> public_location_prefix=http://people.ubuntu.com/~robertc/
> push_location_strip_prefix=hostname:public_html/
> 
> Lastly, it should do a push for us I think :).
> 
> Rob

The current locations that I check for a public branch are:

Config: public_branch = foo
Branch.get_push_location()
Branch.get_parent()
Branch.base

I currently don't have any way of setting 'public_branch' other than
editing the config file yourself. But that is why I was soliciting UI
feedback.

I thought about having it do 'bzr push' before we submit to the pqm. I
haven't decided what I wanted.

I also considered having some sort of string replacement for the
'public' location. So that you could translate 'sftp://foo/path' into
'http://baz/altpath'

I've also looked into the pqm code, and I think I have a minimally
invasive way of adding rio requests. (Basically I change the per-line
with a queue of available lines, it also cleans up the 'patch'
functionality).

By the way, with pqm the only thing that could break out of a running
'patch' command is another 'patch' command. Which ends up completely
overwriting the old 'patch'. Was there a specific design here, or was it
just a mistake in the code?

John
=:->

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060223/eeac057e/attachment.pgp