Merge tests

[Matt Mackall]

On Sun, 2010-05-02 at 09:19 -0400, Greg Ward wrote:
> Hi all --
> 
> I'm reading through test-merge*, trying to find a good place to test
> the various "can't merge with ancestor" cases since I want to improve
> the error messages.  Some of the older ones are a bit confusing:
> 
>   * test-merge2: appears to be testing the merge with working dir case
> when files are removed, but that's unclear. A comment would be nice.

The history claims it's about simultaneous deletion:

> Here's an hg changeset patch that deals with simultaneous deletion of
a
file
> in both the working directory and in a merged branch.
>
> Test case included in the patch.

>   * test-merge3: contains no "merge" or "update" commands, so it can't
> possibly be doing a merge.  (Right?)

Right. From the commit message:

Here's an hg changeset that deals with "hg add"ing and then "hg remove"ing
a file without ever committing it.  Previously, this caused a python
exception
at commit time.

> Some of the other test-merge[0-9] scripts are also a little
> underdocumented, although they're not too hard to figure out with a
> little hg log'ing.  It would be nice to have comments right in the
> scripts, though.  So if some can explain what's up with
> test-merge{2,3}, I'll see if I can add some useful comments.
> 
> Incidentally, I don't see a good place to test the "abort: can't merge
> with ancestor" errors, so I guess I'll add a new test script.

I'm skeptical. Any merge test should have ancestors in it that you can
attempt to merge with.

-- 
Mathematics is the supreme nostalgia of our time.