Re: new snapshot available: coreutils-8.20.119-54cdb0

[Assaf Gordon]

Strange failure with numfmt on an eccentric system (Mac OS X 10.6.8): some 
errors are not reported correctly.

===
FAIL: tests/misc/numfmt
=======================

< ... snip ...>
numfmt.pl: test fmt-err-3: stderr mismatch, comparing fmt-err-3.1 (expected) 
and fmt-err-3.E (actual)
*** fmt-err-3.1 Mon Feb 11 11:55:16 2013
--- fmt-err-3.E Mon Feb 11 11:55:16 2013
***************
*** 1 ****
! numfmt: format 'hello%' ends in %
--- 1 ----
! numfmt: invalid format 'hello%' (width overflow)
fmt-err-4...
numfmt.pl: test fmt-err-4: stderr mismatch, comparing fmt-err-4.1 (expected) 
and fmt-err-4.E (actual)
*** fmt-err-4.1 Mon Feb 11 11:55:16 2013
--- fmt-err-4.E Mon Feb 11 11:55:16 2013
***************
*** 1 ****
! numfmt: invalid format '%d', directive must be %['][-][N]f
--- 1 ----
! numfmt: invalid format '%d' (width overflow)
fmt-err-5...
fmt-err-6...
numfmt.pl: test fmt-err-6: stderr mismatch, comparing fmt-err-6.1 (expected) 
and fmt-err-6.E (actual)
*** fmt-err-6.1 Mon Feb 11 11:55:16 2013
--- fmt-err-6.E Mon Feb 11 11:55:16 2013
***************
*** 1 ****
! numfmt: format '%f %f' has too many % directives
--- 1 ----
! numfmt: invalid format '%f %f' (width overflow)
fmt-err-7...
fmt-err-8...
fmt-err-9...
fmt-err-10...
numfmt.pl: test fmt-err-10: stderr mismatch, comparing fmt-err-10.1 (expected) 
and fmt-err-10.E (actual)
*** fmt-err-10.1        Mon Feb 11 11:55:16 2013
--- fmt-err-10.E        Mon Feb 11 11:55:16 2013
***************
*** 1 ****
! numfmt: grouping cannot be combined with --to
--- 1 ----
! numfmt: invalid format '%\'f' (width overflow)
fmt-err-11...
numfmt.pl: test fmt-err-11 failed: exit status mismatch:  expected 0, got 1
fmt-1...
numfmt.pl: test fmt-1 failed: exit status mismatch:  expected 0, got 1
fmt-2...
numfmt.pl: test fmt-2 failed: exit status mismatch:  expected 0, got 1
fmt-3...
numfmt.pl: test fmt-3 failed: exit status mismatch:  expected 0, got 1
fmt-4...
numfmt.pl: test fmt-4 failed: exit status mismatch:  expected 0, got 1
===

It looks like all the errors are related to the "--format" option, and even 
more strangely, most do produce errors but the wrong error messages.

One example is:

    # On a 'normal' system:
    $ ./src/numfmt --format "hello%"
    ./src/numfmt: format 'hello%' ends in %

    # On this broken system
    $ ./src/numfmt --format "hello%"
    ./src/numfmt: invalid format 'hello%' (width overflow)

In the source code, it seems to be related to this part, in 
parse_format_string(), line 972:
 970   i += strspn (fmt + i, " ");
 971   errno = 0;
 972   pad = strtol (fmt + i, &endptr, 10);
 973   if (errno != 0) 
 974     error (EXIT_FAILURE, 0,
 975            _("invalid format %s (width overflow)"), quote (fmt));
 976            

On this system (Mac OS X):
   fmt = 'hello%'
   i = 6
   fmt+i = ''

And 'strtol' returns errno=EINVAL (22) instead of 0 - causing the incorrect 
error message.

Investigating further...

 -gordon

P.S.
This is a strange system, already has two other bugs associated with it (#13516 
#13495).
Perhaps it's not worth supporting it?