Re: OT: Programming Languages (Was: Spell Checking for Non European Languages, and for Conlangs)

----- Original Message -----
From: "the storyteller" <taliesin-conlang@...>
To: <CONLANG@...>
Sent: Thursday, April 01, 2004 12:05 PM
Subject: Re: [CONLANG] OT: Programming Languages (Was: Spell Checking for
Non European Languages, and for Conlangs)


> * And Rosta said on 2004-04-01 00:03:07 +0200
> > > I wrote:
> > > > Being a syntactician and not a programmer, I wonder why
> > > > programmers don't use trees [..]
[...]
> > I meant trees not as a mathematical structure but as a graphical
> > notation device, contrasting with bracketing and indentation, the
> > opposing merits of which were being debated in the passage I had been
> > responding to. Syntacticians would never use indentation or bracketing
> > for complex structures, since experience shows that they're much
> > harder to read than tree notation.
taliesin:
> Ah, but how many tens of thousands of branches per single tree do the
> average syntactitican operate with at a time? For small things, I agree,
> trees are better, but for larger programs you aren't going to remember
> all of the great monster anyway, instead keeping a vague notion of the
> whole, the tiny spot you're currently working on and its immediate
> context (callers, users etc.).
Philippe Caquant:
> But going back to program packages: how would you use
> a tree to represent a subroutine which is called from
> many different programs ? Clearly, such a subroutine
> couldn't belong to a single branch of the tree. And
> what about recurrent functions ?
I don't see how bracketing or indentation would help in such cases
either.

John Cowan:
> There are code editors like that, that allow you to insert scaffolding
> (an if-then-else construction, e.g.) as a single action, and then lets
> you expand and contract to fill it in.  Experience shows that while such
> editors are very good for examining and modifying programs locally, they
> don't work well for creating programs de novo or for making substantial
> modifications.  The easiest way from a well-formed syntactic form to a
> different well-formed syntactic form is often through something quite
> ill-formed.
>
> In addition, one must be certain that the grammar of the tool exactly
> matches the grammar of the programming language; it's hard to keep the
> two in step.
I'll take your word for it, but it still surprises me that programmers,
of all people, should be content to code by means of such primitive
graphical methods.

> > Syntacticians would never use indentation or bracketing for complex
> > structures, since experience shows that they're much harder to read
> > than tree notation.
>
> But when you publish them, you probably scribble some boxes and arrows on
> paper, and let the poor typesetter figure out what you meant.  Coders have
> no such luxuries.
Oh I know. Nowadays one can't get away with scribbling by hand; everything
is word-processed. But I do give a lot of thought to how to avoid drawing
trees, both because of the labour involved and the page space they take
up (what with articles and chapters tending to have a max page limit).
But I generally -- but perhaps naively -- suppose programmers to be the
ablest of the able and the cleverest of the clever (along with
philosophers -- or maybe they're just the wisest of the wise), and so
the most likely to have insisted on finding a solution for such petty
hindrances.

--And.

Reply