[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: macros vs. blocks

>>>>> "Anton" == Anton van Straaten <anton@appsolutions.com> writes:

Anton> Michael Sperber wrote:
>> While everything else you say is true, SSAX:make-parser is a poor
>> example for an intrinsic use of macros: it's a macro purely for
>> efficiency reasons (at least it was the last time I looked at it) and
>> has a trivial counterpart in procedural form.

Anton> You could have put it a bit more gently!

Sorry.  I'm German, after all :-)

Anton> Whose side are you on?  ;)

I'm on the side of using macros only where there's significant
benefit.  I do agree there are a number of situations where there is
significant benefit.

Anton> Performance is one of the purposes of the macro in this case, but it's not
Anton> the only one.  The SSAX:make-parser macro takes up to seven named callbacks
Anton> as arguments.  Not all of the callbacks have to be supplied; defaults will
Anton> be used for those that are omitted.  The call looks something like:

Anton> (SSAX:make-parser
Anton>   NEW-LEVEL-SEED (lambda (...) ...)
Anton>   UNDECL-ROOT (lambda (...) ...))

Anton> ...where the capitalized names name the subsequent callback procedure.

Anton> This design serves at least two purposes, which I'll describe before
Anton> comparing to the alternatives:

Anton> First, it provides a convenient interface for constructing parsers.  The
Anton> host language here, Scheme, doesn't intrinsically have named arguments to
Anton> procedures, but that feature has been created for this macro, and used to
Anton> provide three of the usual benefits of named arguments: the ability to omit
Anton> arguments, to list them in arbitrary order, and the improved readability
Anton> that can come from explicitly naming them.  Together, this allows a complex
Anton> parser to be constructed quite flexibly, with a single expression, while
Anton> retaining comprehensibility.

Well, the alternative is simply saying:

  'NEW-LEVEL-SEED (lambda (...) ...)
  'UNDECL-ROOT (lambda (...) ...))

This doesn't look all that bad to me.  (In fact, I did just that for
the original PLT/Scheme 48 port of SSAX which I did in April or so, so
I know that it works.  It's possible to get rid of even that by
DEFINEing the keywords to something.  Some Lisp/Scheme dialects
implicitly define identifiers starting with a : to symbols with the
same name which also solves the problem.)

So I think macros purely for the purpose of quote-sprinkling seems
like overkill to me.  If applied too liberally (as in the case of the
scsh shell subsystem, for example), it can lead to confusing programs.

Anton> Now, on to performance: some people have suggested that this is
Anton> unimportant.

All the performance gains in this case could be gained from a
reasonable inlining compiler.  I don't know if Oleg's done any
measurements, but the original motivation for making SSAX:maker-parser
came from making it run faster in the Gambit-C *interpreter*.

Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla