[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Monotone-devel] more on commit without -b option
From: |
Derek Scherger |
Subject: |
[Monotone-devel] more on commit without -b option |
Date: |
Sun, 11 Nov 2007 20:48:48 -0700 |
User-agent: |
Thunderbird 2.0.0.6 (X11/20071004) |
After a brief chat on irc today I was left with the following
impressions as to where this might go.
- remove -b from commit (Thomas Keller already has this done)
- add a mtn branch command for creating new branches that sets the
workspace branch option for future commits. This should fail if the
new branch already exists (Thomas has this done too although I'm not
sure it fails if the target branch exists).
- remove the -b option from update and require that update stay on the
workspace's current branch (i.e. fail if the specified rev is *not* on
the current branch)
- add an mtn switch command that switches the workspace to the specified
branch. Optionally allow -r to specify a particular revision in the
target branch. Require the target branch to exist or fail.
- see what other commands we can remove the --branch option from
- see if we can remove make_branch_sticky and friends
There's a few things in the net.venge.monotone.commit_without_-b branch
that stand out.
- a diff between h:net.venge.monotone and
h:net.venge.monotone.commit_without_-b seems to show a few things that
shouldn't be there like changes to automate common ancestors and the
addition of database::get_common_ancestors, I'm not sure what that's
about
- I like the idea that the branch command with no arguments lists the
current branch
- could we remove the branch guessing code from commit? I can't think of
a case where saying commit without first having set the branch makes
sense. (maybe we could just do this on mainline?)
- (exceedingly minor nit) there's a "change" to
tests/(imp)_deleting_directories/__driver__.lua that adds a single
space to an otherwise empty first line which should be removed
Thoughts?
Cheers,
Derek