[patch] fix bug 48136 bzr status, diff, etc do not work properly after remote push

[John Arbash Meinel]

Aaron Bentley wrote:
> John Arbash Meinel wrote:
>>> Cheuksan Edward Wang wrote:


...

>>> Sorry this took so long to review, +1 from me. Aaron, can you give a
>>> second +1?
>>>
>>> Also, I wonder if this is only part of the fix. Because we want heavy
>>> checkouts to act like lightweight ones, we probably should also be
>>> checking if tree.last_revision() matches (pseudocode)
>>> tree.branch.master().last_revision(). But I'll let that be a follow up
>>> patch. For one thing, it adds a potential network connection to
>>> something that could otherwise be really fast. (bzr status in a heavy
>>> checkout).
> 
> Yes, that's been the problem I've had when considering this before.
> 
> I'm not sure we really want this in diff.  Diff can potentially use
> all-local data, even when the branch is remote.
> 
> Using heavyweight checkouts instead of lightweight ones with remote
> branches is just a convention.  So doing this does require a connection
> to the branch, which is a new requirement.
> 
> Doing it in status seems more sensible, but I don't know whether it
> should be a warning.  Wouldn't it make sense to include it as part of
> the normal status output?
> 
> So I'm at +0.5 here.  I don't want diff doing this.  I don't require
> changing the way it appears in status, but it's at least worth considering.
> 
> Aaron

Well, at present because of the way 'BranchReference' is defined, we
always connect to the remote branch, for any lightweight checkout. I've
thought about having the connection be deferred until we actually use
it. Also, if we have any '--revision' specifier, then we still have to
connect to the remote branch.

I would be okay with it as part of the regular status information, but
I'm not sure where to put it.

I'm also trying to get our patch queue unstuck. We have a lot of stuff
that sits around because it only solves 80% of the problems. But the
problems aren't solved at all without it. So if we can come up with an
alternative in a couple days, I'm willing to wait. I don't really want
to wait another release cycle, though.

So what do you think of:

% bzr status
added:
  foo
  bar/
modified:
  baz
conflicts:
  Text conflict in bling
pending merges:
  Joe Foo 2006-10-27 Message from beyond
info:
  Local tree is out of date with branch, use 'bzr update'

It kind of seems that a warning might be nicer than having it with the
other text. But if we can figure out where to put it, I'd be fine with
it there instead.

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/20061027/ae5deada/attachment.pgp