Re: What's cooking in git.git (Jul 2010, #03; Fri, 16)

From: Elijah Newren <newren@gmail.com>
Date: Jul 17, 2010, 16:24 UTC
Message-ID: <AANLkTimCVgxc9TEujm9ih1fRm6UKUcRwD6DNf59bZgSh@mail.gmail.com>
In-Reply-To: <7vaapqr6a7.fsf@alter.siamese.dyndns.org>

Hi,

On Sat, Jul 17, 2010 at 12:06 AM, Junio C Hamano <gitster@pobox.com> wrote:

Show 5 quoted lines

> * en/fast-export-fix (2010-07-07) 2 commits
>  . fast-export: Add a --full-tree option
>  . fast-export: Fix dropping of files with --import-marks and path limiting
>
> Seems to have breakage in test 9350 (possibly others but I didn't look).

The patch series works on top of maint or master, but not on top of next or pu. The existing tests pass, but the new one it adds breaks because (1) this new testcase depends on the full output of fast-export (and in particular the order of printed filenames) and (2) the "fix output order of D/F changes" from the below series (which underwent some modifications after the original submission of the above series) changes that order.

Show 8 quoted lines

> * en/d-f-conflict-fix (2010-07-09) 6 commits
>  (merged to 'next' on 2010-07-14 at 2b2a810)
>  + fast-import: Improve robustness when D->F changes provided in wrong order
>  + fast-export: Fix output order of D/F changes
>  + merge_recursive: Fix renames across paths below D/F conflicts
>  + merge-recursive: Fix D/F conflicts
>  + Add a rename + D/F conflict testcase
>  + Add additional testcases for D/F conflicts

One fix is fairly trivial -- just change the testcase to use the newer file output order. Alternatively, I could modify the new test to not depend on the full output of fast-export. I'll do the latter and resubmit.

Sorry about this problem; I should have retested the combination of the two series after modifying one of them.

Elijah