Re: Generating loops

Raphael 'kena' Poss wrote:
> Op 21 sep 2010, om 12:53 heeft Gary het volgende geschreven:
>
>> I should say, I'm not married to this syntax. If something else is
>> easier to process then as long as it is readable as a "for" construct
>> I'm not fussed.
>
> It looks like you are tempted to use M4 to implement a compiler of
> sorts.

Err, no :) I am (atually, "we are" - I'm not the only one) just
unfortunate enough to sometimes have to write code in a language which
isn't fit to call itself a language. Source to source translation is as
much as I want, and only for a couple of small things like missing loop
constructs (having to sit and manually write them using those gotos is
just a PITA).

I looked at antlr but given the scope of what I want to do it was
complete overkill in terms of the work involved. At that point I wanted
to be able to write C - so I could not only write code in a meaningful
language but also have some tool support as a result - and generate the
target script code from that; I haven't abandoned that idea entirely,
but kind of need something that will give me results more quickly, even
if the results are less far reaching.

The other stuff I posted about over the last couple of days is a
completely different subject, I just see an opportunity to unify the
tools used.

> Actually I am involved a project which more-or-less does what you are
> doing (the syntax to solve your problem could be "for(i, 1, 12)"
> instead of "for i = 1 to 12" like in C)

That would work. Actually I was hoping to be able to redefine whatever
"," is called to be " " or something (like one can with comments and
quote marks).

> 1) as your substitution rules become more and more complex

It's not *my* language. I'm offended! ;) But "toy" is being generous. Or
do you not mean the target language?

> (you *will* want to add more and more features to your toy language),

If you mean the target language, I can't. If you mean the "language"
that m4 would translate... well, you might be right :">

> 2) there are transformations that are non-trivial to perform with M4;
> say for example you want to define a syntax like:
>
>     PREFIX(f, g, h) code... SUFFIX
>
> and you want this to expand to:
>
>     f(code...)  g(code...)  h(code...)
>
> or anything of the kind where the same piece of input code is
> processed multiple times in different ways, your M4 code will quickly
> become unreadable.

Unreadable? I can't even write it :))