Re: [Apache-dispatch-devel] Fwd: SF.net SVN: apache-dispatch: [50]trunk

[Fred M. <fr...@ta...>]

On Wed, 30 Aug 2006, Geoffrey Young wrote:
>> I've tried an install here on a different system and it barked a bit about
>> couldn't finding Apache::Test but then built properly.  I built
>> Apache::SizeLimit trunk in the same conditions and it refused to build
>> because it couldn't find a mod_perl enabled Apache.
>
> hmm, that's odd.

I'll spin it up again and report back to Philip.

>> Any thoughts on these test results?
>
> they seem ok :)
>
> when I gave it a runthrough I met this problem when testing against mp1
>
> [geoff@spinnaker Apache-Dispatch-svn]$ perl Makefile.PL -httpd
> /usr/local/apache/bin/httpd  && make && make test

Ah good stuff.  I tried something like this on my powerbook at work and 
encountered a different error because my stock httpd was not mod_perl 
enabled, but then I wondered 'how many people will this really affect', 
and I think you've answered that.

> [   info] generating script t/TEST
> Using mod_perl/2.000003

....

> waiting 60 seconds for server to start: .Syntax error on line 18 of
> /src/Apache-Dispatch-svn/t/conf/extra.last.conf:
> Can't locate loadable object for module Apache::Dispatch in @INC
>
> so, the problem is that I didn't use MOD_PERL=1 in my build args.  which
> I kinda remembered about, but I'm pretty sure a random user would be
> very confused by the message - the mp2 line above was my only clue after
> staring at the error for a few minues.

I agree, it's not informative.  I guess what I'm still wondering is since 
a random user would most likely install from CPAN, what is their setup 
going to be like?  Will they have a mod_perl enabled Apache in their path? 
My guess is the random user will not, but instead it's built somewhere 
else - could be wrong though.  An expert user like you will use the -httpd 
option so it sounds like we want to make sure to hit both those use cases.

> so you might want to think about that a bit.  but other than that I
> think sufficient work has gone into this that you should feel to release
> whenever the moment strikes you.  as long as you keep the current CPAN
> version available in case things go awry, there's no real reason to hold
> off I'd think.  and, you should be rewarded for all your hard work :)

I'll hit this again this week and see if I can fix the error 
you ran into.  There's a lot of possible build cases here, it seems like 
this is a common situation for mp1/mp2 Apache::* modules but there is no 
common build solution that fits all yet.  Thanks for your help testing 
this Geoff, I appreciate it :)