Friday 27 April 2012 A love letter to Boost.Spirit

From:  	"Jez Higgins" 
Date:  	Thu, January 4, 2007 11:29 am
To:  	"Kevlin Henney" 

Hi Kevlin,

Alongside the normal keynotes, sessions, panel discussions and so on
that conferences usually have, the academic conferences I went to as a
fresh faced young thing also had poster presentations.  A poster
presentation was, as you might have guessed, a poster of usually about
A3 size, which you filled with whatever you fancied.  The posters were
usually displayed on freestanding boards in the mingling/coffee drinking
areas between the main session rooms.  The nice things about doing a
poster were it was easier (because you could present a small piece of
work, or work in progress), it was cheaper (because you didn't actually
have to be there), but you could still list it as a publication.

I appreciate this is rather late in the day, but wondered if BoostCon
might consider posters?  I only bring it up because I doubt I can afford
to attend, it clashes with my wife's birthday in any case, but I do have
the scrap of an idea for a poster.  It would be some little extracts I'd
written about Boost.Spirit while working on an XPath and XSLT engine,
printed in 72pt text.

   XPath: variables returning node-sets
   ------------------------------------
   Made a minor grammar fix, and reworked location steps slightly so
   expressions like $doc/step1/step2 or $doc[pred] work (assuming $doc
   is bound to a node-set obviously).

   Everytime I find a problem in the grammar, which actually isn't very
   often at all, I give thanks to the pragmatic programming Gods that I
   used Spirit rather than rolling my own parser. In this case, I'd
   transcribed
     FilterExpr ::= PrimaryExpr | FilterExpr Predicate
   after eliminating right-recursion as
     FilterExpr = PrimaryExpr >> *(Predicate >> PrimaryExpr);
   rather than
     FilterExpr = PrimaryExpr >> *Predicate;.

   Time to fix - about 30 seconds. Time to fix if I'd hand-rolled -
   don't even want to think about it :)

   Jaxen Bug 94 - http://jira.codehaus.org/browse/JAXEN-94
   --------------------------------------------------------
   "Charles Souillard reports a problem while trying to evaluate the
  following expression : 2+1-1+1. It should be 3.0 but is 1.0"

   Jaxen is an XPath engine that uses a hand-built parser.  It turned
   out that evaluation order of the arithmetic operators was wrong, and
   the code had to be patched in many different places to fix it.  Since
   my code implements the grammar as written in the spec, I knew that
   not only did I not have this problem, I *could* not have this
   problem.

   When is an XPath expression not an XPath expression?
   ----------------------------------------------------
   When it's an XSLT pattern, that's when.

   As an XPath
     para[@ref]
   selects, for a given node, its child para elements which have ref
   attributes. As an XSLT pattern it tests whether a given node is a
   para element which has a ref attribute.

   Similarly, something like
     chapter/para
   tests whether a given node is a para element with a chapter element
   parent.  In XPath terms, you'd express this as something like
     boolean(self::para[parent::chapter]).

   Aside from this sematic difference, syntactically XSLT patterns are
   an XPath subset. I don't have these things evaluating yet but, thanks
   to the miracle that is Boost.Spirit, I've extended Arabica::XPath to
   parse them. Took about half an hour, including writing a pile of test
   cases and getting them passing.  Spirit really is a top piece of
   work. While it's not hard to find people rambling on about how Python
   or Lisp or whatever really is the thing for writing little domain
   specific languages, this kind of thing really shows just how deadly
   cool template metaprogramming can be. Read the EBNF, transcribe it
   into code, compile, and go. Easy, expressive and type-safe. What more
   could you want?

Yes, it's a love letter to Spirit.  Why not?

Hope you're keeping as busy as you'd like, and that you and your family
are well and happy.
Jez

---

Tagged code, boost, and c++

---