Re: make disttest issues

2003-03-24 Thread Michael G Schwern
On Sun, Mar 23, 2003 at 09:19:04AM -0600, Ken Williams wrote:
   anyway, I noticed the other day that the disttest target seems to 
 have an issue - at least for one of the ways that Apache::Test lets 
 you use it.  given
 
   $ perl Makefile.PL -apxs /usr/local/apache2/bin/apxs
 
   disttest leaves off the additional configuration stuff and just 
 calls perl Makefile.PL.
 
 I wasn't aware that 'make disttest' was supposed to add any other 
 arguments to its 'perl Makefile.PL'.  Is that documented somewhere?

Ahh, of course!  disttest, simulating what happens when you distribute
your module, shouldn't pass through your local command line options. 
Unless someone has a convincing argument otherwise.


-- 
Pancakes is the better part of valor.
http://www.goats.com/archive/971202.html


Re: make disttest issues

2003-03-24 Thread Geoffrey Young

I wasn't aware that 'make disttest' was supposed to add any other 
arguments to its 'perl Makefile.PL'.  Is that documented somewhere?


Ahh, of course!  disttest, simulating what happens when you distribute
your module, shouldn't pass through your local command line options. 
Unless someone has a convincing argument otherwise.
well, I can see both sides and I'm not going to argue (too much) in my favor 
here :)

however, what I use disttest for is to verify that things like my MANIFEST 
are current - basically I compare how the test looks when run with disttest 
(in addition to making sure it passes, of course).  so, in the case that 
brought it to my attention, disttest couldn't run at all because it was 
missing a crucial bit of configuration information.

so, yeah, I guess that disttest as it works now lets me know what happens 
when somebody doesn't configure the Makefile properly.  but it doesn't 
really let me know what an end-user will see when they do configure it 
properly.  of course, I can test the former myself by mis-configuring the 
initial Makefile, but I have no easy way to test the latter without some 
additional support.

anyway, thanks for taking the time to at least think about it :)

--Geoff



Re: make disttest issues

2003-03-24 Thread Geoffrey Young
Ideally, Makefile.PLs should be able to run without any extra args or user
interaction taking at least sensible defaults.  Shouldn't it be able to
scan around a few likely locations for your Apache directory?  At the
very least, you can fall back to an environment variable.
yeah, the appropriate routines do fallback to an environment variable, so 
yes there are ways for me to get around the issue.  there may not be for 
others, though, which is what I was trying to say (I think... it was early).

in general, it was just a suggestion.  I think I've heard enough to 
understand the issues and see that I'm probably using disttest in ways it 
wasn't intended - after all, if nobody has complained about this by now, I 
guess it's not important to the masses.

thanks

--Geoff



Re: make disttest issues

2003-03-23 Thread Michael G Schwern
On Fri, Mar 21, 2003 at 03:26:26PM -0500, Geoffrey Young wrote:
   anyway, I noticed the other day that the disttest target seems to have an 
 issue - at least for one of the ways that Apache::Test lets you use it.  
 given
 
   $ perl Makefile.PL -apxs /usr/local/apache2/bin/apxs
 
   disttest leaves off the additional configuration stuff and just calls 
 perl Makefile.PL.  the relevant lines in my target are
 
 disttest : distdir
 cd $(DISTVNAME)  $(ABSPERLRUN) Makefile.PL
 
 anyway, I guess the routines in Apache::Test that parse additional 
 command-line arguments are non-standard, but it would be nice to support 
 this feature, especially since there are other targets that seem to store 
 this information just fine.

Post up your Makefile and maybe we can figure out what's missing.


-- 
1. It Has To Work.
-- RFC 1925


Re: make disttest issues

2003-03-23 Thread Ken Williams
On Friday, March 21, 2003, at 02:26  PM, Geoffrey Young wrote:

hi all...

  I didn't realize that such a list existed, but it's a much better 
place to post my, um, feature report than p5p :)  appologies if I'm 
posting to the wrong place.

  anyway, I noticed the other day that the disttest target seems to 
have an issue - at least for one of the ways that Apache::Test lets 
you use it.  given

  $ perl Makefile.PL -apxs /usr/local/apache2/bin/apxs

  disttest leaves off the additional configuration stuff and just 
calls perl Makefile.PL.
I wasn't aware that 'make disttest' was supposed to add any other 
arguments to its 'perl Makefile.PL'.  Is that documented somewhere?

Seems like maybe what you want to do is just 'make test', unless I'm 
misunderstanding.

anyway, I guess the routines in Apache::Test that parse additional 
command-line arguments are non-standard, but it would be nice to 
support this feature, especially since there are other targets that 
seem to store this information just fine.
I think it's pretty common for Makefile.PLs to take other command-line 
options.

 -Ken



Re: make disttest issues

2003-03-23 Thread Ken Williams
On Sunday, March 23, 2003, at 02:17  PM, Michael G Schwern wrote:

On Sun, Mar 23, 2003 at 09:19:04AM -0600, Ken Williams wrote:
 anyway, I noticed the other day that the disttest target seems to
have an issue - at least for one of the ways that Apache::Test lets
you use it.  given
 $ perl Makefile.PL -apxs /usr/local/apache2/bin/apxs

 disttest leaves off the additional configuration stuff and just
calls perl Makefile.PL.
I wasn't aware that 'make disttest' was supposed to add any other
arguments to its 'perl Makefile.PL'.  Is that documented somewhere?
Ahh, of course!  disttest, simulating what happens when you distribute
your module, shouldn't pass through your local command line options.
Unless someone has a convincing argument otherwise.
Yeah, that's what I was thinking.  It's supposed to be similar to an 
automatic smoke test or the like.

 -Ken