Randy W. Sims | 2 Oct 2005 07:05

Re: [Module::Build] Solving the common-MANIFEST.SKIP entries problem

Ken Williams wrote:
> 
> On Sep 28, 2005, at 5:40 PM, Randy W. Sims wrote:
> 
>> Michael G Schwern wrote:
>>
>>> B)  Get 0.28 out the door, damnit!  I want my fixed installbase! [1]
>>> [1] Consider this to be a gentle prod towards release.
>>
>>
>> I finally got some relief at work; this weekend will be the first, in 
>> many months, that I haven't had to work. I had planned to celebrate 
>> and take it easy this weekend, but I'll try to go ahead and clean up 
>> the remaining issues that I introduced or have been involved in, 
>> excluding the integration of the new argument processing code.
> 
> 
> I think the only issues we need to clear up before the next beta are 
> currently-failing tests.  There's one that I added to t/moduleinfo.t for 
> another $VERSION-parsing thing that it's currently failing on, I think 
> we could just make that a TODO item.

Fixed.

> There's a test failure that I'm 
> seeing on Cygwin in t/destinations.t that should really get fixed, 
> though, since I think it's likely to actually bite people.  I've been 
> looking at it but I don't understand it yet (I unfortunately can't spend 
> much time on the Win32 box to narrow it down), so any help from anyone 
> (not just Randy S.) would be great.
(Continue reading)

Ken Williams | 3 Oct 2005 02:55
Favicon

Re: [Module::Build] M::B does not always respect its own verbose/quiet flag


On Sep 30, 2005, at 11:23 AM, Dave Rolsky wrote:
>
> Oh, I just hadn't checked.  Also, in my patch before this I sent some 
> new tests for moduleinfo.t.  You fixed the bug differently than in my 
> patch but never added those tests.  I think it'd be handy to have 
> those in there.
>

It looked like those tests were meant to be in their own test script, 
so I had added them as t/metadata2.t , hoping we could get them patched 
into an existing test script instead.  If you could submit such a patch 
that would grease the wheels a little bit.

  -Ken

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Ken Williams | 3 Oct 2005 03:49
Favicon

Re: [Module::Build] Solving the common-MANIFEST.SKIP entries problem


On Oct 2, 2005, at 12:05 AM, Randy W. Sims wrote:

> Ken Williams wrote:
>> I think the only issues we need to clear up before the next beta are 
>> currently-failing tests.  There's one that I added to t/moduleinfo.t 
>> for another $VERSION-parsing thing that it's currently failing on, I 
>> think we could just make that a TODO item.
>
> Fixed.

Excellent.

>
>    File::Spec->catdir( File::Spec->rootdir, 'bin', File::Spec->curdir 
> );
> => //bin/
>
>    File::Spec->catdir( File::Spec->rootdir, File::Spec->curdir, 'bin' 
> );
> => //./bin
>
> The last example works; it removes the trailing slash

Weird - File::Spec on Cygwin is just a very thin wrapper around the 
Unix version, which doesn't seem to have this problem.

>
> This is with both the default 3.05 and the newer 3.11. Both File::Spec 
> and Module::Build will need fixes...
(Continue reading)

Ken Williams | 3 Oct 2005 03:58
Favicon

Re: [Module::Build] Abbreviating output


On Sep 30, 2005, at 5:38 AM, David Golden wrote:

> If the implied question is, should this be changed, I'd vote "yes, but 
> in 0.30".  Probably at the same time that all the "print" statements 
> are changed over log_warn, log_info, etc. and the verbosity handled 
> there.

Heck, I'll take a patch for either issue at any time.  Neither will 
require coordinating with anything else, AFAIK.

>
> Higher on my wishlist is figuring out why the %*&$ 
> CPANPLUS/Dist::Build has *no* output for many types of failures.  
> Makes debugging my FAIL's rather hard.

I do know why this is - it's because CPANPLUS runs M::B through its 
API, not as a subprocess, and since Test::Harness sends this output 
straight to STDERR/STDOUT, it goes uncaptured.

It occurs to me that now that CP+::D::B is under our control, though, 
we should just be able to patch it there.

The best place to patch would probably be right in Test::Harness, but 
of course that doesn't help people who still have the old versions.

  -Ken

-------------------------------------------------------
This SF.Net email is sponsored by:
(Continue reading)

Randy W. Sims | 3 Oct 2005 05:16

Re: [Module::Build] Solving the common-MANIFEST.SKIP entries problem

Ken Williams wrote:
> 
> On Oct 2, 2005, at 12:05 AM, Randy W. Sims wrote:

>>    File::Spec->catdir( File::Spec->rootdir, 'bin', File::Spec->curdir );
>> => //bin/
>>
>>    File::Spec->catdir( File::Spec->rootdir, File::Spec->curdir, 'bin' );
>> => //./bin
>>
>> The last example works; it removes the trailing slash
> 
> 
> Weird - File::Spec on Cygwin is just a very thin wrapper around the Unix 
> version, which doesn't seem to have this problem.
> 
> 
>> This is with both the default 3.05 and the newer 3.11. Both File::Spec 
>> and Module::Build will need fixes...
> 
> 
> True, it would be a good fix, but I wonder if we could fix in the M::B 
> tests by just running the paths through canonpath()?

I played with this for a while last night and couldd not get anything 
that would work from File::Spec. I tried canonpath, rel2abs, etc. 
Anytime I try to turn it into a absolute path, it insists on adding an 
extra slash at the beginning and the end of the path. I'm thinking of 
just comitting a hack like:

(Continue reading)

Randy W. Sims | 3 Oct 2005 09:40

[Module::Build] [PATCH] ExtUtils::CBuilder - process defines & extra_compiler_flags

Module::Build in cvs now depends on this updated version 0.15 to handle 
correct processing of defines & extra_compiler_flags.

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Dave Rolsky | 3 Oct 2005 19:19
Gravatar

Re: [Module::Build] M::B does not always respect its own verbose/quiet flag

On Sun, 2 Oct 2005, Ken Williams wrote:

> On Sep 30, 2005, at 11:23 AM, Dave Rolsky wrote:
>> 
>> Oh, I just hadn't checked.  Also, in my patch before this I sent some new 
>> tests for moduleinfo.t.  You fixed the bug differently than in my patch but 
>> never added those tests.  I think it'd be handy to have those in there.
>
> It looked like those tests were meant to be in their own test script, so I 
> had added them as t/metadata2.t , hoping we could get them patched into an 
> existing test script instead.  If you could submit such a patch that would 
> grease the wheels a little bit.

Hmm, I thought the patch I sent was for moduleinfo.t.  Looking at 
metadata2.t, I don't see those tests.  But looking at modulinfo.t I see 
you _did_ add them, just in a slightly different form.  Sorry for the 
confusion.

-dave

/*===================================================
VegGuide.Org                        www.BookIRead.com
Your guide to all that's veg.       My book blog
===================================================*/

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
(Continue reading)

Dave Rolsky | 3 Oct 2005 20:08
Gravatar

[Module::Build] More $VERSION parsing fun

So I found a module on CPAN which does this:

( $VERSION ) = ...

The current ModuleInfo doesn't parse this, so I fixed it and added some 
tests for this style.  I also added /x to the version regexp and commented 
it as it was getting quite hairy.

-dave

/*===================================================
VegGuide.Org                        www.BookIRead.com
Your guide to all that's veg.       My book blog
===================================================*/

Index: lib/Module/Build/ModuleInfo.pm
===================================================================
RCS file: /cvsroot/module-build/Module-Build/lib/Module/Build/ModuleInfo.pm,v
retrieving revision 1.11
diff -u -r1.11 ModuleInfo.pm
--- lib/Module/Build/ModuleInfo.pm	2 Oct 2005 02:28:27 -0000	1.11
+++ lib/Module/Build/ModuleInfo.pm	3 Oct 2005 18:07:46 -0000
 <at>  <at>  -11,9 +11,19  <at>  <at> 

  my $PKG_REGEXP  = qr/^[\s\{;]*package\s+([\w:]+)/;
-#my $VERS_REGEXP = qr/([\$*])(([\w\:\']*)\bVERSION)\b\s*=[^=]/;
-my $VERS_REGEXP = qr/([\$*])(((?:::|')?(?:\w+(?:::|'))*)?VERSION)\b\s*=[^=]/;
-
+my $VERS_REGEXP = qr/(?:\(\s*)?    # $VERSION could be in parens
+                     ([\$*])       # sigil - $ or *
(Continue reading)

Ken Williams | 3 Oct 2005 20:32
Favicon

Re: [Module::Build] M::B does not always respect its own verbose/quiet flag


On Oct 3, 2005, at 12:19 PM, Dave Rolsky wrote:
>
> Hmm, I thought the patch I sent was for moduleinfo.t.  Looking at 
> metadata2.t, I don't see those tests.

Oops, that was David Golden that sent those tests, not you.

If someone could write some kind of tool for managing all the Davids 
and Randys, that would be cool.

  -Ken

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Ken Williams | 3 Oct 2005 20:29
Favicon

Re: [Module::Build] [PATCH] ExtUtils::CBuilder - process defines & extra_compiler_flags


On Oct 3, 2005, at 2:40 AM, Randy W. Sims wrote:

> Module::Build in cvs now depends on this updated version 0.15 to 
> handle correct processing of defines & extra_compiler_flags.

I think you forgot the patch here.

  -Ken

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl

Gmane