If SGML is so great...

[Gary Preckshot <garrell@inreach.com>]

A lot of people say that the reason it's so good is it can
be converted to virtually anything else. Then they launch
into a religious war for their favorite format.

All of this is irrelevant. If it can be converted into
anything, then one honker of a machine (owned by an LDP
sponsor) can take all the SGML and turn it into the format
de jour. It's a non issue.

Likewise, if SGML can be translated into anything else, the
reverse translation can be done into a subset of SGML. There
isn't any reason for LDP to require SGML. Anything that can
be translated into SGML should be acceptable. LDP can use
SGML internally, but there isn't any justifiable reason for
imposing it or DocBook on HOWTO authors. This is the year
2000 after all, and beloved 486s can be replaced by machines
with 100 times the memory, disk, and speed for $1500. Just
because some authors can't afford it is no reason to limit
those who can and will.

I'm really puzzled by the stick-in-the-mud attitude of some
of the people on this forum. It's possible to cater to the
people with older machines AND advance the ability of LDP to
accommodate modern formats. All that's really necessary is
to standardize internally, and build translators to go from
popular formats to the internal format. It's even good
design, because if a successor to SGML comes along, the
transition can be internal, and the author community won't
even see it. After all, the pace of improvement in memory,
disk, and processor speed is a factor of 2 every 18 months
or so, and I replaced my 33 Mhz 486 three years ago with a
233 Mhz machine with 64 Mb of memory and (now) 60 Gb of
disk. The cost of a similar machine is now less than $1000,
and the cutting edge is around 800 Mhz with 128 Mb of RAM
and 70 Gb of disk. All for about $2500. Sometimes it seems
the LDP is replaying the IDE disk mess, fighting the wars of
30 years ago over issues that died 30 years ago.

There really isn't any need for these irrelevant arguments.
Sure, SGML is wonderful. But some of us just want to write a
simple mini-HOWTO that doesn't use all the bells and
whistles. Why should we be forced to learn yet another page
description language when all we want to do is write? We're
not limited by cretaceous era 486s, and we can run
translators with no sweat. Why not. Why should we be limited
to the capabilities of the slowest 486 in the LDP universe?
If all we need is a subset of SGML produced by automatic
translators, so be it. I'm not trying to start a flame war,
but just trying to expand the horizons of the LDP. If LDP is
to compete in the greater arena, it has to accommodate
authors of diverse abilities and resources. Including those
who don't care about the wonders of SGML. There's no room
for religious arguments about which format prints the nicest
or whatever. The objective of LDP is Linux documentation,
not page description languages; they're only a tool.

I'm going to look into WP9. If I never have to deal with
SGML, it's ok with me. I don't care about it, I don't want
to have to deal with it, and my computer is here to do the
grunt work, not me. First there was machine language, which
got tiresome fast. Then there was assembly language, which
still gets tiresome. Then there was C. I was there for all
of it. Now there are higher level languages than C. SGML is
the assembly language of document preparation. I'd like to
use the C of document preparation at least.

Gary


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