git testsuite failures

git testsuite failures

[David S. Miller]

There are test cases failing in the testsuite over the
past few days, is this expected?

*** t6001-rev-list-merge-order.sh ***
 ...
* FAIL 3: simple merge order check_output simple-merge-order git-rev-list --merge-order --show-breaks HEAD
* FAIL 4: two diamonds merge order (g6) check_output two-diamonds-merge-order-g6 git-rev-list --merge-order --show-breaks g4
 ...
* FAIL 8: cross-epoch, head at l5, prune at l1 check_output cross-epoch-head-at-l5-prune-at-l1 git-rev-list --merge-order l5 ^l1
* FAIL 9: duplicated head arguments check_output duplicated-head-arguments git-rev-list --merge-order l5 l5 ^l1
 ...
* FAIL 20: max-count 10 - merge order check_output max-count-10-merge-order git-rev-list --merge-order --show-breaks --max-count=10 l5
 ...
* FAIL 23: --max-age=c3, --merge-order check_output max-age-c3-merge-order git-rev-list --merge-order --max-age=51148811 l5
 ...

Re: git testsuite failures

[Jon Seymour]

I posted a fix for this yesterday. The break was caused by a change in
behaviour of parse_commit that the --merge-order stuff was dependent
on. Now that parse_commit is sane, the --merge-order stuff needs to be
made sane too.

The fix was in this patch.

[PATCH 1/1] Fix --merge-order unit test breaks introduced by
6c88be169881c9223532796bd225e79afaa115e1

jon.

Re: git testsuite failures

[Linus Torvalds]

On Wed, 22 Jun 2005, David S. Miller wrote:
> 
> There are test cases failing in the testsuite over the
> past few days, is this expected?

Fixed now. It was the parent ordering cleanup that broke an assumption
about parents being in reverse order in the rev-list "epoch" code. I
checked in the fix from Jon,

		Linus