git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 0/3] solaris test results
@ 2008-02-20 23:59 Jeff King
  2008-02-21  0:00 ` [PATCH 1/3] git_config_*: don't assume we are parsing a config file Jeff King
                   ` (3 more replies)
  0 siblings, 4 replies; 28+ messages in thread
From: Jeff King @ 2008-02-20 23:59 UTC (permalink / raw)
  To: Junio C Hamano; +Cc: Whit Armstrong, git

I have now run every test script <= t9001 using the current 'master' on
the SunOS 5.8 box available to me. It _mostly_ passed. Following are
three patches:

  1/3: git_config_*: don't assume we are parsing a config file

       This is an actual minor bug which was covered by most platforms'
       ability to print (null) for NULL. I posted this earlier, but with
       a typo.

  2/3: t3404: use configured shell instead of /bin/sh

       This is a problem purely in the test script, not git itself.

  3/3: diff: fix java funcname pattern for solaris

       This is a Solaris-specific bug in git.

With these fixes, everything passes for me with a few exceptions:

  - Sun's diff doesn't understand "-u". I was able to use GNU diff.
    Since comparing actual and expected output is so common, we could
    potentially abstract this with a "test_cmp()" function and use
    something platform specific. It's probably not worth the trouble, as
    it impacts only the test suite, and only on systems with a totally
    broken diff.

  - t4020 fails without gnu grep, as it requires "grep -a"

  - t4024 fails unless /usr/xpg4/bin is in your path, as the regular
    version of 'expr' is unhappy with us (I didn't investigate too
    closely)

  - t4118, t4200, t8001, and t8002 fail unless you have GNU sed. The
    xpg4 sed barfs if there's no newline at the end of input (and this
    is specifically part of these tests)

  - I couldn't personally run t5502, t5505, t5512, t5520, or t9001, all
    of which require the Git.pm module; the version of perl on this
    machine is old (5.005_02) and complained about the 'our' keyword

  - t7505 uses a hook which hard-codes /bin/sh but uses syntax that
    the Solaris /bin/sh doesn't like. Maybe hooks should get their
    $SHELL value replaced?

  - t3900 mostly passes, but fails to correctly convert between EUCJP
    and ISO-2022-JP (in either direction). I didn't investigate very
    far.

  - git-filter-branch invokes "sh" for the commit filter, which invokes
    the broken Solaris /bin/sh. This isn't wrong, exactly, but it might
    make more sense to invoke $SHELL.

Now I must go gouge out my eyes, so that I may never see the output of
Solaris /bin/sh ever again.

-Peff

^ permalink raw reply	[flat|nested] 28+ messages in thread

end of thread, other threads:[~2008-02-24 19:41 UTC | newest]

Thread overview: 28+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2008-02-20 23:59 [PATCH 0/3] solaris test results Jeff King
2008-02-21  0:00 ` [PATCH 1/3] git_config_*: don't assume we are parsing a config file Jeff King
2008-02-21  0:00 ` [PATCH 2/3] t3404: use configured shell instead of /bin/sh Jeff King
2008-02-23 20:13   ` Johannes Sixt
2008-02-23 20:26     ` Junio C Hamano
2008-02-23 20:34     ` Junio C Hamano
2008-02-23 21:09       ` Johannes Sixt
2008-02-23 21:15         ` Jeff King
2008-02-23 21:56           ` Junio C Hamano
2008-02-23 22:39             ` Jeff King
2008-02-23 22:46               ` Junio C Hamano
2008-02-23 22:49                 ` Jeff King
2008-02-23 22:55                   ` Junio C Hamano
2008-02-23 23:12                     ` Jeff King
2008-02-24  0:03                       ` Junio C Hamano
2008-02-24 19:40                         ` Jeff King
2008-02-23 20:40     ` Jeff King
2008-02-23 20:50       ` Junio C Hamano
2008-02-23 21:12         ` Jeff King
2008-02-23 22:28           ` Jeff King
2008-02-23 22:35             ` Junio C Hamano
2008-02-23 22:44               ` Jeff King
2008-02-21  0:01 ` [PATCH 3/3] diff: fix java funcname pattern for solaris Jeff King
2008-02-21  0:34 ` [PATCH 0/3] solaris test results Junio C Hamano
2008-02-21  0:41   ` Jeff King
2008-02-22  5:26     ` Junio C Hamano
2008-02-22  5:29       ` Junio C Hamano
2008-02-22  5:42       ` Jeff King

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).