[tpm] Bullet Points from Tonight's TPM Presentation

James E Keenan jkeen at verizon.net
Thu Aug 27 20:23:32 PDT 2009


Live and raw from New York City to Toronto ...

Perl/Parrot Miscellany:  Talk at Toronto Perlmongers, Thursday,  
August 27, 2009, 7:00 pm

I.  Parrot Virtual Machine:
* distribute copies of Allison's article

* emphasize success of monthly release program, with major releases  
in Jan (n.0) and Jul (n.6) of each year
** 1.4 theme:  stable interface for language implementers
** 2.0 theme:  production

* Parrot Developer's Guide:  PIR, by Allison Randal and Andrew  
Whitworth, now available:
** http://www.powells.com/biblio?isbn=0977920127

* What can we use:
** tests on uncommon platforms -- particularly if you can help with  
debugging on those platforms
*** we still have open RTs from Parrot Build Fest at TPM in March 2008:
**** http://rt.perl.org/rt3/Ticket/Display.html?id=52204
**** http://rt.perl.org/rt3/Ticket/Display.html?id=52212

** help from someone who understands linking on Mac OS X particularly  
re OpenGL and SDL
*** https://trac.parrot.org/parrot/ticket/924
*** http://rt.perl.org/rt3/Ticket/Display.html?id=52212

** noobs:  read http://docs.parrot.org and tell us what doesn't make  
sense


II.  Rakudo implementation of Perl 6 on Parrot Virtual Machine
* distribute copies of pmichaud's use.perl.org posts

* pumpking's request:  Write programs in Perl 6 and give Perl 6 team  
feedback


III.  Perl 5 Corehackers Project
* promoted by Chip Salzenburg at YAPC::NA::2009 in Pittsburgh
** was anyone there for the closing speech?

* a license to hack on both the Perl 5 core (C) and core modules

* xdg requested my assistance with ExtUtils::ParseXS
** a module which is little more than a wrapper around the xsubpp  
program written originally by Larry Wall
** doesn't work under 'use strict'; a model of Perl Worst Practices;  
but it works for what it does
** let's get it to run under 'use strict' and be better tested and  
more maintainable
** we'll need both blackbox and whitebox testing
*** blackbox:  Does any given revision of ParseXS.pm produce the same  
C output for a given input (modulo whitespace) -- where "given input"  
means "any .XS file currently in CPAN?  Create a minicpan; identify  
all .XS files; understand how they're called by xsubpp, i.e., with  
what options; write script that compares "before" and "after" versions
*** whitebox:  phalanx it

** It's also apparent that 'perlxstut' is inadequate.  At least one  
example only works on Linux (not even on Darwin).   Other examples  
may be broken entirely.

** ParseXS is inadequately tested.  We need to write dummy.XS files  
which display all of the different possibilities for parsing.  To get  
there, we should go through 'perlxs' and annotate each paragraph with  
a test specification.
*** Example:  In the test suite there shall be a dummy.XS file that  
displays:  SCOPE: ENABLED
** Once we have a complete test specification, then we (or the more  
XS-knowledgeable subset of 'we') can knock out the tests.


More information about the toronto-pm mailing list