Re: [PATCH] Disable filemode rebase tests in t3400-rebase.sh if no filesystem mode support

From: Alex Riesen <raa.lkml@gmail.com>
Date: Feb 11, 2009, 11:05 UTC
Message-ID: <81b0412b0902110305o5ca60708gc07cbc9ae1d89d1f@mail.gmail.com>
In-Reply-To: <7v7i3xgu9v.fsf@gitster.siamese.dyndns.org>

2009/2/11 Junio C Hamano <gitster@pobox.com>:

Show 14 quoted lines

> Alex Riesen <raa.lkml@gmail.com> writes:
>
>> Signed-off-by: Alex Riesen <raa.lkml@gmail.com>
>> ---
>>
>> I'm honestly sorry for my contributions. They seem to be exclusively
>> in the "make Git work in Windows" area.  Depressing.
>>
>>  t/t3400-rebase.sh |    5 +++++
>>  1 files changed, 5 insertions(+), 0 deletions(-)
>
> I wonder if you should be fixing the tests with "update-index --chmod"
> instead...
>

Well, mostly these should work independent of the filesystem (except for the cases when the index is checked out later). They seem to, and they look like they should, too. See t4014-format-patch, for one.

But you're right. I should have reviewed them earlier. Anyway, there is not many tests which have "--chmod" in them: t4006-diff-mode.sh, t4014-format-patch.sh, t6031-merge-recursive.sh, and t9500-gitweb-standalone-no-errors.sh. The t4006 does not bother to look at working tree (good), t4014 is good (looks like its ignoring modes altogether), t6031 uses --chmod to augment the chmod useless on FAT-like fs (seem to be an earlier fix for the same problem).

What's more interesting is the use of chmod for index mode change purpose, which should be reviewed regularly (and I do that only when it breaks. Lazy boom, I am...)