Re: What's cooking in git.git (Mar 2011, #04; Wed, 23)

Junio C Hamano [off-list ref] writes:

I won't be doing the tests, though.  At least not tonight.

If anybody is interested in tackling this while I am sleeping (gee, it is
almost 1am now and I am not so young anymore to pull an all-nighter),
the minimum set of things to test are as follows:

 * with no configuration, "git merge" errors out;
 * with configuration set to "false", ditto;
 * with configuration set to "true":

   - on detached HEAD, "git merge" errors out;

   - without branch.*.remote, ditto;

   - without branch.*.merge, ditto;

   - with branch.*.remote and branch.*.merge set, but without
     remote.<value of branch.*.remote>.fetch that specifies RHS
     to store the fetched value, "git merge" errors out;

   - does it work correctly when branch.*.remote, branch.*.merge and
     remote.<value of branch.*.remote>.fetch are all set reasonably?

   - does it work correctly when branch.*.remote is ".", i.e., forked
     from a local branch?

   - does it work correctly when more than one branch.*.merge are
     set for the current branch, i.e. configured to always create an
     octopus?

 * does it have any funny interaction with various command line options
   that are eaten by parse_options()?