Front page | perl.perl5.porters |
Postings from October 1999
Re: [p5p] Re: SimpleDoc (SDF successor) request for input
Thread Previous
|
Thread Next
From:
Russ Allbery
Date:
October 7, 1999 01:51
Subject:
Re: [p5p] Re: SimpleDoc (SDF successor) request for input
Message ID:
ylvh8j8sfn.fsf@windlord.stanford.edu
Brad Appleton <bradapp@enteract.com> writes:
> On Tue, Sep 28, 1999 at 09:06:02AM -0700, Russ Allbery wrote:
>> My understanding is that SDF is attempting to solve a larger problem
>> and is therefore semantically richer
> This part is very much is correct.
>> sufficiently so that it's no longer "plain."
> This part is very much incorrect however.
> SDF is about as "plain" as it gets. IMHO its much plainer than POD
> *most* of the time. The only time it isn't is for the more
> sophisticated/advanced cases where POD really can't even do what you
> require. There are a few areas where SDF's markup looks sufficiently
> different from PODs as to make each camp think the other's markup is the
> less plain of the two; but those are really six of one and half dozen of
> the other.
Well, for what it's worth, I really don't agree with this, and I did look
at how SDF worked once upon a time so the conclusion isn't completely
baseless. I went and re-read some of the basics now, and it still strikes
me the same way. It's quite possible that you and I have different
definitions of plain, though.
My primary complaint about SDF as a language is that it makes far more
extensive use of special characters than I'm comfortable with. POD is
very simple and easy to explain largely because there are only three
special escapes in the entire language: a paragraph beginning with =, a
paragraph indented with whitespace, and something of the form X<>.
To this, SDF adds [[ ]], !, :, [], >, << and >> in some circumstances, and
with lists the whole barrage of *, -, ., ^, +, and &. I find punctuation
inherently unintuitive unless you've used it a lot, unavoidably so, so
there's a strong correlation between more punctuation and more complexity
as far as I'm concerned.
I'm not claiming that = or X<> are intuitive (I would argue that indenting
something with whitespace actually is to some degree), but that's a pretty
short list of things to have to learn.
So my argument is that the *syntax* of SDF is no longer "plain." The
appearance may very well be, perhaps even more so than POD for the case of
lists, but I can't apply the same very straightforward parsing tactics
that I can apply to POD and there's a wider *variety* of visual clutter.
I'm willing to learn a lot of punctuation for my programming language, but
I really like documentation formats that are as simple as possible. The
only reason why I don't just write everything in plain text is that it's
worth avoiding ambiguity in a few places that are hard to avoid in
completely unstructured plain text.
I would rather expand POD in those places where it needs some expansion (I
was realizing today that there are times when I'd really like to have a
=quote command that behaved like an HTML <blockquote> tag; I wonder if one
could just use =over/=back for that). Maybe if I felt a strong need for
POD to support tables, I'd be more interested in seeing it merge with SDF
and pick up SDF's features, but I honestly haven't seen a need for that
yet.
Anyway, it's probably worth keeping in mind that I'm perhaps something of
a dinosaur on the subject of markup and text-processing languages, having
acquired a lot of fairly strongly-held prejudices over the years. I've
been writing in text processing languages for about seventeen years now,
and to a large degree the entire direction in which text processing as a
whole is going (XML) is anathema to me. Syntaxes that I consider to be
the worst of all available worlds are winning the mindshare battle for The
Way To Do Text Processing. So you all may not want to listen to me, as I
seem to be on the losing side. *wry grin*
In the meantime, I'll probably continue working to support very plain POD
as an input format and translating it directly to text, *roff, and the
like; I'm not interested in writing XML, and while I appreciate its
importance, I'd rather do a really good job of turning POD into text than
plug half-heartedly at figuring out how to build XML DTDs.
--
Russ Allbery (rra@stanford.edu) <URL:http://www.eyrie.org/~eagle/>
Thread Previous
|
Thread Next