hg-1.4.1 test, solaris: 2 errors left, test-hook, test-bisect

[Mads Kiilerich]

rupert THURNER wrote, On 12/15/2009 07:39 PM:
> hi,
>
> with the hg-1.4.1 release the test result got again better, down from 5
> errors to now two. great, and many thanks!
>
> $ TESTFLAGS="-j $(psrinfo | wc -l)"
> $ time ./run-tests.py $TESTFLAGS
> +++ Test output
> @@ -305,7 +305,8 @@
>
>   % test invalid command
>   /bin/sh: ./foobar: not found
> -abort: failed to execute ./foobar
> +Changeset 5:7874a09ea728: bad
> +abort: cannot bisect (no known good revisions)
>   % test bisecting command
>   Testing changeset 15:e7fa0811edb0 (31 changesets remaining, ~4 tests)
>   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
> !........s...s..........s...............

Do you have any idea how we should handle this when we expect

$ foo ; echo $?
dash: foo: not found
127

but with Solaris /bin/sh we get

$ foo ; echo $?
foo: not found
1

Should we just remove that part of the test? Or skip the test on 
solaris? Or accept that if fails on solaris? Or the remove the 127 = 
"command not found" check from bisect?

/Mads