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

Re: Editors, Searching, Geekness, and Gary Preckshot

Peter Hernberg wrote:
> First: Editors
> Gary seems to have some sort of super-huge issue with text editors, especially
> vi.

Nope. I have a huge issue with using inadequate tools.
DocBook is complicated. Writing DocBook with a text editor
is an unnecessary test of your virtuosity. BTW, I use emacs
on X-Windows. It's an awful editor in text format, and I
actually use vi when I'm in Unix (or Linux) and in text
mode. You really only need to know 4 commands to use vi, and
that's all I know. Vi is adequate for text editing, but I
use X and emacs if I'm programming. I haven't tried the
psgml feature, but I've been told it has the requisite
feature of adaptable tag lists.
> Second: Searching
> Gary's issue with editors, in some vague sgml-works-in-mysterious-ways sort of
> thing, is that he believes that searching problems are caused by users of text
> editors.

Nope. We don't have any searching problems because we don't
have a search engine.

> Gary Preckshot doesn't know a thing about sgml searching.

Speak for yourself.

> Neither do I.

I believe you.

> That's the point! SGML searching involves first buiding an sgml parser (from
> what I understand, a very difficult task) and then doing something intelligent
> with the parsed data (getting computers to do intelligent things can be
> difficult ;).

lesson 1: It's not an SGML parser, it's a DocBook (or other
DTD) parser.
lesson 2: Garbage in, garbage out. If DocBook isn't used in
a way that leaves data to search for, there won't be
anything to search for.
lesson 3: I've been making computers do difficult things for
30 years.

> Gary, unless you've been doing a very good job hiding your
> intricate knowledge of sgml parsers and searching, don't make a big deal about
> it.

DocBook can be parsed with a simple push-down stack
recognizer. The syntax is simple. See Aho.

> I haven't and won't say much about searching because I don't know much about
> it.

Yep. You said it. I didn't. However, I agree with you.

> If you want to learn something about the subject, you might crack open the
> third volume of Donald E. Knuth's The Art of Computer Programming. If you can
> even understand the first three sections, I will forever bow to your immense
> knowledge of searching.

See Aho, Sethi, and Ullman. Knuth is good for algorithms,
but these three wrote the book on compilers (or parsers, for
the uninitiated).
> Third: Geekness
> I am a self-acknowledged geek.

Same here, but I do something about it. I recognize the
difficulty in communicating with normal people, and take
steps to facilitate it.

SGML is complicated. It's product (DTDs and document
description languages) is not. Goldfarb invented it because
he was tired of redoing content every time he changed
presentation. As it happened, what he invented was a way of
tagging content so that it could be rendered by rule based
rendering engines. What he actually invented was a flexible
way of representing diverse content in a way that was
amenable to database techniques. The world was not slow to
recognize this, there being many commercial applications for
this. Most of what you see on the web is SGML based. Every
time you do an on-line order, it's likely you're interacting
with a DTD search engine. Amazon.com is probably a heavy
user. Have a look and ask yourself how they handle GB of
text and images.


To UNSUBSCRIBE, email to ldp-discuss-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org