root/misc/STD_red/README

A ruby transliteration of src/perl6/STD.pm

INSTALL

The release of ruby 1.9.0 (dated late December '07) is recommended.
It will not work with ruby 1.8.  ruby svn HEAD is also not the right thing.
There used to be iffy support for 1.8, but adding utf handling broke it.

http://www.ruby-lang.org/en/news/2007/12/25/ruby-1-9-0-released/
$ ruby --version
ruby 1.9.0 (2007-12-25 revision 14709) [...]

Debian:
 ruby1.9 libreadline-ruby1.9
 * Debian (testing aka "Lenny") error:
   `require': no such file to load -- readline (LoadError) from -e:1:in `<main>'
   Means you don't have libreadline-ruby1.9.


EXAMPLE
  ./STD_red_run -e 42

  time ((find ../../v6/v6-KindaPerl6/t/kp6/ -type f ; find ../../t/ -type f )| grep -v '\.svn' | sort | xargs -n 1 perl -e 'local $_=shift;exit if !/\.t$/;$r=system("./STD_red_run $_ > /dev/null 2>&1");print "",($r == 0 ? "-" : "X")," ",$_,"\n";' > test-status)
  ~5 minutes

NOTES

STD_red_run is currently much faster in ruby 1.9 than 1.8.
If you compile large things, you probably want to make sure its #! line finds 1.9.
Either by having 1.9's "ruby" first in your PATH, or by editing the #! line.

Regex reminders
  given { <a>+ <b>+ }
  token: /^ <a>+: <b>+: $/
  rule:  /^ <.ws> <a>+: <.ws> <b>+: <.ws> $/
  regex: /^ <a>+ <b>+ $/
which transliterate as
  token:  plusTOK{a} and plusTOK{b}
  rule:   wsp and plusTOK{a} and wsp and plusTOK{b} and wsp  #handwritten rules
  rule:           plusTOK{a} and wsp and plusTOK{b}          #in token rules #XXX hmm
  regex:  plusRX(lambda{ a }){ plusRX{ b }}
    # but note, <a> can't be a regex - we won't backtrack into it.

There are lots of regex ruls.  Only the two noted as backtracking in
comments actually do.

Re backtracking,
  plusRX et al, _do not backtrack into their subrules_.
  We are simplifying implementation by noting there are tokens everywhere.
  If there turns out to be a case of a regex with backtracking, containing
  a subrule which is itself a regex with backtracking, then we'll need to
  hand fudge passing a continuation to that subrule.  Very hopefully, the
  case won't arise.
  We're not making a regexp engine, nor a real Grammar.
  We're simply trying to get the ability to parse static p6, by the
  easiest possible development path.


RUBY YAML

Is not being used now, so you can ignore this section.

--yaml won't work under ruby 1.9 without a patch applied to 1.9's yaml.rb.
 *** But --yaml isn't being used now, so this is no longer needed.
Error ruby/1.9.0/yaml.rb:391:in `hash': can't convert Hash into Integer (TypeError)
ruby 1.9.0 yaml has a bug.  On some of Match objects, (eg, --yaml -e '3'),
.hash fails with : can't convert Hash into Integer (TypeError).
But 1.9's lookbehind provides 10x faster parsing than than our current 1.8 workaround.
So here is a patch to lib/ruby/1.9.0/yaml.rb:
PATCH_START
--- yaml.rb.orig        2008-03-20 13:25:42.000000000 -0400
+++ yaml.rb     2008-03-20 13:26:03.000000000 -0400
@@ -386,6 +386,7 @@
             end
         oid =
             case oid when Fixnum, NilClass; oid
+            when Hash,Array; oid.object_id
             else oid = "#{oid.object_id}-#{oid.hash}"
             end
         out.emit( oid, &e )
PATCH_END