perl-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Geoffrey Young <>
Subject Re: svn commit: r419247 - /perl/Apache-SizeLimit/trunk/Makefile.PL
Date Thu, 06 Jul 2006 14:37:36 GMT

>>  my %args = @Apache::TestMM::Argv;
>>  die 'suitable httpd required'
>>    unless ($args{apxs} or $args{httpd} or
> It's too bad this isn't an Apache::Test method/function. That'd be quite
> handy.

yeah, it would.  unfortunately, even if it were it might be too late in
the game for it to be truly useful.

>> where the user is Apache-Test savvy.  I _think_ it also covers when a
>> user has ~/.apache-test and/or, but now I just can't
>> remember.
> Unfortunately it doesn't, which is kind of bad, since I have
> and it's kind of important that I can run the tests.

Apache-Test savvy people can always run the tests

  $ perl Makefile.PL -httpd /path/to/your/httpd

> That may be why I first looked at that code and wanted to remove it.

right.  I was taking the position that it's better to not run the tests
(and risk an improper setup from the unaware) than to cause developers
or in-the-know people an extra step.

>> make sense?
> I'm with you on the goal, but I think it needs to check to for
> and ~/.apache-test as well to be useful.

I guess that depends on your definition of useful.  doing so would make
it run tests for more users automatically, so in that sense it's good.
but doing so would also mean more work for us

> Is there a separate check I can do for those? Looking at
> Apache::TestConfig I cannot figure out how to do it without invoking the
> prompts.

I don't think so.  I think the process needs to be something like
creating the config then checking the test server object.  but creating
the config is done with

  t/TEST -config

during the test cycle.  so what _really_ needs to happen in order to be
effective is creating the config during the 'perl Makefile.PL' phase
rather than the 'make test' phase.  and I think I tried doing that a
while ago but couldn't make it work properly.

but at any rate, I don't think we need to decide this now and hold up
the release any longer.  my thought is that we should allow this release
to focus its attention on the API changes and memory issues that
prompted it, rather than need to wade through bogus bug reports.  so in
that sense, I'd rather users not run the tests if they don't know how to
properly configure their environment.

but in the end, I leave the choice to you :)


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message