Re: [PATCH v7a 6/9] diff --stat: use a maximum of 5/8 for the filename part

From: Jeff King <peff@peff.net>
Date: Mar 26, 2012, 23:45 UTC
Message-ID: <20120326234549.GA10172@sigill.intra.peff.net>
In-Reply-To: <1330604806-30288-6-git-send-email-zbyszek@in.waw.pl>

On Thu, Mar 01, 2012 at 01:26:43PM +0100, Zbigniew Jędrzejewski-Szmek wrote:

Show 5 quoted lines

> ---
>  Documentation/diff-options.txt |   14 ++++---
>  diff.c                         |   90 ++++++++++++++++++++++++++--------------
>  t/t4052-stat-output.sh         |   16 +++----
>  3 files changed, 76 insertions(+), 44 deletions(-)

I am seeing test failures from t4052 in 'master'. Bisecting points to 1b058bc (diff --stat: use a maximum of 5/8 for the filename part, 2012-03-01). The output from the test script looks like this:

--- expect 2012-03-26 23:41:29.688039554 +0000 +++ actual 2012-03-26 23:41:29.696039549 +0000

@@ -1 +1 @@
- ...aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa |    1 +
+ ...aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa |    1 +
not ok - 8 diff: small change with long name gives more space to the name
#       
#                       git $cmd $args >output &&
#                       grep " | " output >actual &&
#                       test_cmp expect actual
#               

There are a few other failures, but they all have the same mismatched
length (the output is slightly longer than expected). I know Junio won't
push out a 'master' that doesn't pass the tests for him, so I'm
wondering if some environment information like the terminal width is
leaking through the test scripts.

I haven't actually investigated any further yet, but I thought I'd first
see if anything obvious occurs to you.

-Peff