Remove 'running-make-test'

[jkeenan]

In the course of a #parrot discussion today of a test failure reported by darbelo, plobsing and I attempted to figure out what t/harness option --running-make-test does.

I grepped the repository and came up with this:

./docs/intro.pod:148:
   C:\Perl\bin\perl.exe t\harness --gc-debug --running-make-test
./lib/Parrot/Harness/Options.pm:31:
      $longopts{running_make_test} = grep { $_ eq '--running-make-test' } @argv;
./lib/Parrot/Harness/Options.pm:32:
      @argv = grep { $_ ne '--running-make-test' } @argv;
./lib/Parrot/Harness/Options.pm:133:
      --running-make-test
./t/pharness/03-handle_long_options.t:25:
      --running-make-test
./t/pharness/03-handle_long_options.t:31:
  ok($longopts->{running_make_test}, "Captured long option");
./t/harness:34:
  $ENV{RUNNING_MAKE_TEST} = $longopts->{running_make_test};
./t/harness:192:=item C<--running-make-test>
./config/gen/makefiles/root.in:1890:
  # The --running-make-test argument is currently used by long-running tests
./config/gen/makefiles/root.in:1892:
  EXTRA_TEST_ARGS := --gc-debug --running-make-test

I couldn't see what difference setting $ENV{RUNNING_MAKE_TEST} made. I conducted an Internet search using a well-known search engine, and only came up with links to Parrot posts -- many by me! (Eeek!)

plobsing++ suggested cutting running-make-test (in all its casings and punctuations) out of the distro post-2.0 and seeing what breaks. To explore that, I have created the no_running_make_test branch in SVN.

Can anyone shed any light on what running-make-test is supposed to be doing? Is it simply an historical artifact?

Thank you very much.

kid51

[jkeenan]

Replying to jkeenan:

plobsing++ suggested cutting running-make-test (in all its casings and punctuations) out of the distro post-2.0 and seeing what breaks. To explore that, I have created the no_running_make_test branch in SVN.

Removing running_make_test had no discernable impact on make test in that branch. See  this Smolder report.

[doughera]

On Tue, 19 Jan 2010, Parrot wrote:

>  In the course of a #parrot discussion today of a test failure reported by
>  darbelo, plobsing and I attempted to figure out what ''t/harness'' option
>  `--running-make-test` does.

It does nothing.  Digging through my old files, I was able to find just 
one use of it in

   parrot-0.5.2/languages/perl5/t/sparse_perlarray.t

In that case, it prepended a "gc_debug 0" to the pasm code, presumably to 
make it run faster.  The other consequence is that it made the test run 
via 'make test' *different* from the test run directly (assuming that the 
gc_debug command actually does anything anymore).  I've always found such 
cases confusing.

My vote is "Get rid of it."

-- 
    Andy Dougherty		doughera@lafayette.edu

[coke]

On Tue, Jan 19, 2010 at 8:24 AM, Andrew Dougherty
<doughera@lafayette.edu> wrote:
> On Tue, 19 Jan 2010, Parrot wrote:
>
>>  In the course of a #parrot discussion today of a test failure reported by
>>  darbelo, plobsing and I attempted to figure out what ''t/harness'' option
>>  `--running-make-test` does.
>
> It does nothing.  Digging through my old files, I was able to find just
> one use of it in
>
>   parrot-0.5.2/languages/perl5/t/sparse_perlarray.t
>
> In that case, it prepended a "gc_debug 0" to the pasm code, presumably to
> make it run faster.  The other consequence is that it made the test run
> via 'make test' *different* from the test run directly (assuming that the
> gc_debug command actually does anything anymore).  I've always found such
> cases confusing.
>
> My vote is "Get rid of it."
>
> --
>    Andy Dougherty              doughera@lafayette.edu
> _______________________________________________
> parrot-tickets mailing list
> parrot-tickets@lists.parrot.org
> http://lists.parrot.org/mailman/listinfo/parrot-tickets
>

+1

-- 
Will "Coke" Coleda

[mikehh]

I ran all tests on the branch no_running_make_test at r43478 and everything PASSed.

no_running_make_test branch - All tests PASS at r43478 (pre/post-config, corevm/coretest, test, fulltest) - Ubuntu 9.10 amd64 (g++ with --optimize)

if anything fulltest ran about 5% faster, although I don't think this is conclusive.

+1

[jkeenan]

no_running_make_test branch merged into trunk at r43546. Prior to merge, passed make test and make codetest. Will keep ticket open for 2-3 days for any further complaints.

Thank you very much.

kid51

[jkeenan]

No complaints recorded. When I subsequently merged in the one_make branch, there was a minor glitch due to the presence of one last usage of 'running-make-test', but that was quickly fixed.

Closing this ticket.

Thank you very much.
kid51