#+TITLE:     Notes and ToDo items for Perl 6 Grammars in Pugs
#+AUTHOR:    Austin Frank
#+EMAIL:     austin.frank@gmail.com
#+LANGUAGE:  en
#+OPTIONS:   H:3 num:nil toc:nil \n:nil @:t ::t |:t ^:t *:t TeX:t

* about .pg files in src/Pugs/Parser/

This file was created in emacs' org-mode.  Feel free to edit it in the
Pugs source tree.  I'll keep an eye on the formatting.  For more
information about org-mode, check out the [[http://staff.science.uva.nl/~dominik/Tools/org/][org home page]].

You can find an html-ized version of this file online at [[http://notes.aufrank.net/development/perl6/p6parsing.html][aufrank's site]].

The *.pg files in this directory contain some initial attempts at
writing Perl 6 Grammars for parsing Perl 6.  These are bare bones
sketches, and are all sorts of wrong.  

If you've got questions about the files that are already started, or
decide to start one of your own and have questions about that, please
feel free to add them here.  Likewise, if you know that some other
part of Pugs needs a particular grammar built, feel free to make some
notes on the requirements here.

If you'd like to start working on writing Grammars to parse Perl 6,
here are some of the things that definitely need attention:

* tasks
** TODO Grammar for hash indexing and slicing
   Hash access has a few different sugared forms, using {}, <>, <<>>,
   and their unicode friends I can't type.  A grammar for hash access
   should:
   + put the matched key into $<key>
   + keep track of the access context (literal quoting, interpolated
     quoting, curly-hash-constructor)
*** references
*** questions
    + How should key names that need to be interpolated be stored in
      the match object?  Right now Capture.pg stores them as the
      concatenation of sigil and variable name.

** TODO Grammar for array indexing and slicing
   Array indexing and slicing has become a lot more interesting as
   well, through the introduction of multi dimensional slicing.  A
   grammar for array access should:
   + parse single element slices (normal array access)
   + parse multiple element slices
   + parse multiple dimension single element and multiple element slices

   I haven't thought much about what the structure of the Match object
   should be for Array indexing.  You probably want to have something
   like (the leading colon is just for formatting)
   : $<dim><0><elem> = 1,2,3 
*** references
*** questions

** TODO Grammar for parsing quoted expressions
   Perl has always had plenty of ways of quoting text.  Perl 6 will be
   no different.  We need to parse literal quotes and interpolating
   quotes in their ascii, unicode, and quotelike-operator forms.

   The Match object resulting from parsing a quoted expression should
   have a $<is_interpolated> flag set based on how the expression was
   quoted, and should also have a $<quoted> (or some better name) that
   contains a string representation of whatever was being quoted.

   You could probably pretty easily move on to parsing string
   concatenations (infix:<~>) from here.
*** references
*** questions

** TODO Grammar for parsing start/end punctuation pairs
   This is a hard one, and so far it has been viciously ignored.  When
   a group is opened with a punctuation character like '(', it should
   only be counted as closed if it sees a ')'.  Right now, (...]
   matches just fine (which is bad).

   This particular issue has been recognized as being pretty
   complicated.  See the references below for the kinds of things
   you'll have to think about.
*** references
    + [[http://colabti.de/irclogger/irclogger_log/perl6?date=2006-05-09,Tue&sel=625#l962][#perl6 discussion on matching grouping characters]]
    + [[http://wall.org/~larry/pspe][Larry's page of Ps/Pe unicode charcter classes]]
    + [[http://www.unicode.org/Public/4.1.0/ucd/BidiMirroring.txt][bidimirroring]]
    + [[ftp://ftp.unicode.org/Public/UNIDATA/UnicodeData.txt][UnicodeData.txt]]
*** questions

** TODO Add Synopsis references (and links where possible) to these tasks
** TODO Find out how to import grammars
   This should be just like use'ing a module, but I don't know where
   in the Syns it's described.  Not in S05 or S12, from what I can
   tell.
** TODO Set $<context> on unpacked parameters
   This should be based on which bracketing characters are used for
   the unpacking: [] = @ context, {} = % context, :() = & context (I
   think).

* Archived Notes

** DONE(<2006-05-18 Thu 02:44>) [,]-ify Capture.pg
   CLOSED: [2006-05-18 Thu 02:44]
   thanks to TimToady++, [,] has taken the place of unary *, and list
   context eager has taken the place of unary **.  These changes need
   to be implemented in the named argument handling of Capture.pg
*** references
    perl svn doc repo, r9216


** DONE(<2006-05-18 Thu 02:44>) Resync Capture.pg with Synopsis 6
   CLOSED: [2006-05-18 Thu 02:44]
   audreyt++ checked in a bunch of changes to make named argument
   parsing simpler.  These need to be implemented.  Off the top of my
   head, this includes not treating 'foo' => bar as a named argument,
   and restricting named arguments to using the fat comma pair
   constructor.
*** refernences
    perl6 svn doc repo, r9222