I felt I should cross-post since I'm trying to infect core python developers
with ReviewBoard goodness.  Original and related threads are at:



 ~ Dan

----- Forwarded message from Dan Buch <daniel.b...@gmail.com> -----

Date: Sat, 10 Jul 2010 13:25:32 -0400
From: Dan Buch <daniel.b...@gmail.com>
To: the-fellowship-of-the-packag...@googlegroups.com
Subject: RBTools 0.2.1alpha needed for Mercurial repos (was Re: reviewboard)

> The names given in your ``[paths]`` section are significant.  This is a
> recent addition to RBTools and deserves some documentation, but I'll
> summarize here.  If, when trying to determine the repository type and
> metadata, ``post-review`` determines the repo to be of type "hg", remote
> paths are checked with the following names (in this order):
>     * "reviewboard"
>     * "origin"
>     * "parent"
>     * "default"


This behavior is only present in 0.2.1, which hasn't *yet* been released.
You'll be wanting to install the alpha 0.2.1 to get desired behavior when
working with Mercurial.  Source distributions available here:


> At one point the debug output of ``post-review`` included information
> about this heuristic, but such output was deemed too verbose and backed
> out.
> That said, this is an example ``[paths]`` section (for your local
> repository's ``.hg/hgrc``) that *should* work with ``post-review``::
>     [paths]
>     default = http://bitbucket.org/meatballhat/distutils2
>     origin = http://bitbucket.org/tarek/distutils2
>     ...
>     zubin = http://bitbucket.org/zubin71/distutils2
>     merwok = http://bitbucket.org/Merwok/distutils2
>     ... other remotes ...
> > 
> > > Also: if you upload the diff by hand, make sure you're not using the
> > > git diff format, as this would cause the server to 500.
> > 
> > This is very bad. Non-git diff are really limited, they can’t describe
> > executable bit change, changes to binary files, renames, whereas git
> > diffs are equivalent to a native changeset and thus much more useful.
> I fully agree.  git-style diffs are waaaay better.  There is a
> server-side fix for this.  I will update the ReviewBoard instance now
> that 1.5RC1 is out, which *supposedly* includes the fix.

----- End forwarded message -----

Attachment: pgpllVVblwosW.pgp
Description: PGP signature

Reply via email to