On Mon, Dec 09, 2002 at 12:51:24PM -0500, [EMAIL PROTECTED] wrote:
> !Does anyone know why it's done that way?  Our version keeps getting
> !out of synch with the MM_Unix version that all other platforms use,
> !and it would be much more maintainable if we could get rid of the
> !override.  I also notice that continuation lines are scrupulously
> !avoided in [.vms]descrip_mms.template; is there any good reason for
> !that?
> 
> That is certainly a pain.  I suspect (although I do not actually
> know) that the assignment of the macros was done that way to
> avoid an allocation limit of some sort (along the lines of the 255
> limit for DCL lines passed into lib$spawn()).  Admittedly macros
> are data structures internal to the build tool and their assignment
> /concatenation build up statements should not have
> to worry about DCL limits (they are not action lines after all).
> Perhaps there was an old version of MMS that needed such spoon
> feeding?  I suspect that Charles Bailey might remember why it
> was done that way.  Isn't it the case that you always use MMK?
> 
> As to possible solutions: why not have the perldepend portion
> of MM_$platform.pm built up from a .PL extractraction by
> miniperl at perl build time?  I suspect the only tricky bit would
> be the maintenance of ExtUtils::MakeMaker outside of the perl
> source build process.  But at least the headache of header listings
> would be lessened.

This is what I hope to do, but first I want to remove the duplication
between the MM_Unix->perldepend and MM_VMS->perldepend, the big list of
header files, which lead to this make format question.

Doing it outside the core isn't any harder, you just use arch/CORE/*.h.


-- 

Michael G. Schwern   <[EMAIL PROTECTED]>    http://www.pobox.com/~schwern/
Perl Quality Assurance      <[EMAIL PROTECTED]>         Kwalitee Is Job One
Anti-cow device.  Don't bother.
        http://www.goats.com/archive/990323.html

Reply via email to