Re: rcs-5.9.4: make check: tests 510 511 603 604 605 900 999 failed on AIX

[Thien-Thi Nguyen]

() Michael Felt <address@hidden>
() Mon, 26 Jan 2015 19:49:00 +0100

   I have always 'remembered' RCS as a directory where the
   comma-v files were stored, so I was not surprised to see RCS
   as a directory.

Normally, "RCS" is a directory containing comma-v files.  RCS
should be able to operate just fine with the comma-v in the same
directory as the working file, however.  For most of the tests,
"x" is the working file and "x,v" is the comma-v file (all in
the same tNNN.d directory).

I'm beginning to think that specifying ‘--disable-threads’ is
not the right approach, and that propagating the makefile var
‘LIBTHREAD’ as determined by the configure script, is.  Perhaps
something in gnulib is not so flexible and was not fully tested
w/ threads disabled under AIX.  So, i've just installed:

 http://git.savannah.gnu.org/cgit/rcs.git/commit/?h=p&id=cdc017067
 http://git.savannah.gnu.org/cgit/rcs.git/commit/?h=p&id=6dd5ca388

which do that for "make" and "make check" flows, respectively.
Could you please apply these patches to the freshly-unpacked
Makefile.in files in src/ and tests/, respectively:

Makefile.in.diff
Description: patch for src/Makefile.in

Makefile.in.diff
Description: patch for tests/Makefile.in

then configure (*without* ‘--disable-threads’) and "make" and
"make check VERBOSE=1"?

   What puzzles me though is that the same errors are reported -
   but/and there are err files in test-dirs that passed.

   address@hidden:[/data/test/rcs-5.9.4]make VERBOSE=1 check
   [...]

That's not a problem.  For most tests, the test infrastructure
splits stdout/stderr unconditionally (in tests/common, see shell
func ‘split_std_out_err’).

   On Mon, Jan 26, 2015 at 3:13 PM, Thien-Thi Nguyen <address@hidden> wrote:
   [...]

Could you please omit this bottom-citing in future posts?
We prefer to avoid top-posting on GNU mailing lists, generally.
TIA for your cooperation on this.

-- 
Thien-Thi Nguyen
   GPG key: 4C807502
   (if you're human and you know it)
      read my lisp: (responsep (questions 'technical)
                               (not (via 'mailing-list)))
                     => nil

signature.asc
Description: PGP signature