Re: ExtUtils:MM_Unix v1.33 manifypods patch

2002-11-08 Thread Michael G Schwern
On Thu, Oct 31, 2002 at 03:49:41PM +, Steve Purkis wrote:
> Hi Michael & co.,
> 
> I've been working on a manifypod -related system, and have had to tweak
> ExtUtils::MM_Unix subs so that:
> (a) I understood what they did (is this an obfustacted perl 
> contest? :-)
> (b) I could get my work done.
> 
> My solution lives in a Makefile.PL, but it didn't take long to push it 
> back
> into ExtUtils::MM_Unix and whip up a patch.  Unfortunately I don't have
> time to test it on any other flavour of Unix than Mac OS X 10.1.5 (with
> Perl 5.8 and Gnu Make 3.79), or even to write tests for it - sorry in
> advance.
> 
> But still, I thought you might find it of some use in your quest to 
> clean up
> MakeMaker ;-).  So the patch is attached.

I sort of see what you're doing here, this is just a cleanup, right?  No new
features introduced?

If I could ask you to take another whack a this, modify your patch so that
rather than using the pod2man binary it uses the Pod::Man module. [1]  If
nothing else it'll be a lot faster to have one process calling
Pod::Man->parse_from_file multiple times than calling pod2man once for each
individual POD file. [1]


PS  You might find these slides from the Refactoring talk useful or at least
amusing.  They talk specificly about the icky POD2MAN code.
http://www.pobox.com/~schwern/talks/Refactoring/slides/slide020.html

[1] This means 5.5.3 users will have to install Pod::Man.  Oh well.

[2] 5.6.1 introduced a pod2man which could do multiple conversions in one
command.  Unfortunately, we much be backwards compatible to 5.5.3.

-- 

Michael G. Schwern   <[EMAIL PROTECTED]>http://www.pobox.com/~schwern/
Perl Quality Assurance  <[EMAIL PROTECTED]> Kwalitee Is Job One
Only mindless violence can raise my spirits now!



Re: INSTALLSCRIPT doesn't have 'site', 'vendor' values?

2002-11-08 Thread Ken Williams

On Saturday, November 9, 2002, at 11:36  AM, Michael G Schwern wrote:


On Thu, Nov 07, 2002 at 07:54:18PM +1100, Ken Williams wrote:

If I were to fix this, I think there would be a few parts to it:

 1) MakeMaker should try to determine a better place to put scripts (by
guessing)


Should be possible using the same tricks as the other iffy variables in
init_INSTALL().


Thanks, I'll have a look.


 2) EU::Install should honor that


It'll do that automaticly.  It doesn't know anything about the INSTALL*
variables, it just gets fed paths.  ExtUtils::Install is really just a
glorified copy command like GNU install.


Oh yeah.  Good.


 3) Config.pm should have INSTALLSITESCRIPT and INSTALLVENDORSCRIPT
entries so MM doesn't have to guess


We'll still have to guess for all older versions, so no code savings 
there.
:(

RIght, I meant more to increase predictability and reduce guessing, not 
to actually simplify code.

 -Ken



Re: INSTALLSCRIPT doesn't have 'site', 'vendor' values?

2002-11-08 Thread Michael G Schwern
On Thu, Nov 07, 2002 at 07:54:18PM +1100, Ken Williams wrote:
> It sounds like the kind of thing someone's going to be using somewhere, 
> so it probably shouldn't get completely broken.

The two likely candidates are AFS users, who routinely shuffle files around
all over the place and use obscure features of MM, and module repackagers
(ie. Redhat, Debian, etc...) who need to seperate out architecture dependent
vs independent bits.


> If I were to fix this, I think there would be a few parts to it:
> 
>  1) MakeMaker should try to determine a better place to put scripts (by 
> guessing)

Should be possible using the same tricks as the other iffy variables in
init_INSTALL().


>  2) EU::Install should honor that

It'll do that automaticly.  It doesn't know anything about the INSTALL*
variables, it just gets fed paths.  ExtUtils::Install is really just a
glorified copy command like GNU install.


>  3) Config.pm should have INSTALLSITESCRIPT and INSTALLVENDORSCRIPT 
> entries so MM doesn't have to guess

We'll still have to guess for all older versions, so no code savings there.
:(


> In practice, I don't think any 'hints' files (in 5.8, anyway) have 
> INSTALLBIN and INSTALLSCRIPT different, so it's probably pretty safe to 
> guess from INSTALL*BIN, I think.
> 
> Then I have to figure out how to cooperate/repeat with Module::Build.


-- 

Michael G. Schwern   <[EMAIL PROTECTED]>http://www.pobox.com/~schwern/
Perl Quality Assurance  <[EMAIL PROTECTED]> Kwalitee Is Job One
You see, in this world there's two kinds of people.  Those with loaded
guns, and those who dig.  Dig.
-- Blondie, "The Good, The Bad And The Ugly"



Re: INSTALLSCRIPT doesn't have 'site', 'vendor' values?

2002-11-08 Thread Ken Williams

On Thursday, November 7, 2002, at 05:56  PM, Andreas J. Koenig wrote:


On Thu, 7 Nov 2002 13:18:17 +1100, Ken Williams 
<[EMAIL PROTECTED]> said:


Okay.  I'll wait a week or so for someone like Andreas or Andy to
chime in and explain why there are no other INSTALL*SCRIPT entries,
and if that doesn't happen I'll assume it's an error and I'll try to
fix it.


Sorry, I have no answer to this one.


Would it be wrong to just put scripts into INSTALL*BIN?  Do we really
need to make a big distinction between executable binaries and
executable scripts?


The reason for 'script' being separated from 'bin' was that you can
share scripts between architectures but usually cannot share binaries.

If people actually make use of this distiction, nobody knows.


It sounds like the kind of thing someone's going to be using somewhere, 
so it probably shouldn't get completely broken.

If I were to fix this, I think there would be a few parts to it:

 1) MakeMaker should try to determine a better place to put scripts (by 
guessing)
 2) EU::Install should honor that
 3) Config.pm should have INSTALLSITESCRIPT and INSTALLVENDORSCRIPT 
entries so MM doesn't have to guess

In practice, I don't think any 'hints' files (in 5.8, anyway) have 
INSTALLBIN and INSTALLSCRIPT different, so it's probably pretty safe to 
guess from INSTALL*BIN, I think.

Then I have to figure out how to cooperate/repeat with Module::Build.

 -Ken



Re: All MakeMaker HTMLifying functionality is going away

2002-11-08 Thread Michael G Schwern
On Fri, Nov 08, 2002 at 12:55:01PM -0500, Andy Dougherty wrote:
> We'd give you one if we had a clue what it should be.

Yeah, I don't expect it to be solved.  Besides, the Great Convert Everything
To HTML Craze is mercifully over.

Now let's talk $Config{installyamldir}... ;)


-- 

Michael G. Schwern   <[EMAIL PROTECTED]>http://www.pobox.com/~schwern/
Perl Quality Assurance  <[EMAIL PROTECTED]> Kwalitee Is Job One
"That's what Jagulars always do," said Pooh, much interested.  "They call
'Help! Help!' and then when you look up they drop down on you."



Re: All MakeMaker HTMLifying functionality is going away

2002-11-08 Thread Andy Dougherty
On Thu, 7 Nov 2002, Michael G Schwern wrote:

> That was the reason it was killed.  No way to know the default location for
> HTML docs or when its appropriate to generate and install them.  The problem
> is punted back to the Config folks.  If they can give us a %Config value for
> it then we can use it.

We'd give you one if we had a clue what it should be.  Perl's INSTALL file
has had the following paragraph for years now (i.e. since 5.005_57, in May
1999).

=item HTML pages

Currently, the standard perl installation does not do anything with
HTML documentation, but that may change in the future.  Further, some
add-on modules may wish to install HTML documents.  The html Configure
variables listed above are provided if you wish to specify where such
documents should be placed.  The default is "none", but will likely
eventually change to something useful based on user feedback.

Attempts to get useful user feedback have all failed. 

-- 
Andy Dougherty  [EMAIL PROTECTED]