[
Lists Home |
Date Index |
Thread Index
]
From: "Tim Bray" <tbray@textuality.com>
> etc. It's also a matter of putting in endless hours on ISO committees,
> working with with difficult people in a really exacting process, and
> learning a forest of acronyms and numbers that make the W3C/IETF look
> like child's-play. Volunteers, please step forward.
Quick nurse, the screens! There there Tim, it is only a traumatic memory now :-)
Let me defend SC 34 a little.
There are only a handful of active ISO standards in play at the moment:
DSDL, XTM
not the dozens at W3C.
Then there are a handful of specs which are currently stable or under
incremental maintenance
SGML, DSSSL, ISO HTML, HyTime, the fonts and entities TR
and which really just respond to feedback from users (e.g. XML
being the most high-profile adaptation of an SC 34 standard.)
SC34 is a small, targetted, focussed effort looing after some really key
standards related to "Document Description and Processing Languages".
Jim Mason, the convenor, works really hard to keep it ticking over
smoothly. Its technologies underpin many of the world's largest
industries, and especially some industries of key importance in the
US economy and security.
ISO standards are reviewed every 5 years, and if they
have not found acceptance they are removed. The acceptance is
not based on whether commercial vendors can make a buck
(indeed, it may be the sign of success of a standard that the only
viable implementations are open source and free!), or whether
a standard becomes ubiquitous: it is whether it helps with some
significant problems for some important stakeholders.
"Standards are documented agreements containing technical specifications or
other precise criteria to be used consistently as rules, guidelines, or definitions
of characteristics, to ensure that materials, products, processes and services
are fit for their purpose." [1]
I understand that one of the major aerospace companies generates
about 360 million different (customized-content) pages a year of documentation
from SGML, and this may double over then next couple of years.
Imagine the difficulty of typesetting, printing, collating and distributing this,
let alone creating and maintaining the documents! SGML is probably one of
the most successful standards for that publishing niche, and it is an
immensely important niche.
It is being fit-for-purpose that is important, not whether that purpose is
a mass market or not, or under what brand-name the technology appears.
When we look around for technologies that remind us of SGML circa 1996
-- good for high-end uses, but really needing industry experience to reflect
into a low-end profile for mass uses by the rest of us-- the embarrassed party
is now not ISO but W3C!
I don't see any comparison with W3C on complexity. W3C has far more technology,
and certainly W3C+IETF has. As for an excessive use of acronymns, I guess
Tim is referring to SGML's use of names (such as STAGO: Start
Tag Open) for delimiters ("<"), but this is necessary because they
can be remapped. Things have to have names, as anyone reading
the XQuery group's use of FLWOR, will banefully attest.
Indeed, on complexity, if you just take the ISO 8879 spec (not Goldfarb's
annotated version) it is looking pretty slim compared to XML 1.0 +
XML 1.1 + Namespaces 1.0 + Namespace 1.1 + Infoset 1.0 +
possibly Infoset 1.1 + xml:base + Xinclude + XPath 1 + XML Schemas
+ charmod (which is the full-featured alternative) Trying to claim
W3C specs have the virtue of simplicity is a bit of a joke nowadays,
Tim: you've lost the 80/20 battle, but you are great for continuing to
fight it. Instead, W3C seems to be accepting the "100 minus 20" approach
(big spec, but fob people off who say it is too big by saying "oh, you can subset it")
which is just what was said about SGML.
On Tim's later comment of whether we are all using HyTime, I wonder if
he is contrasting it with XLink's, err, ubiquity? Or DOM AS. Or
Fragment Interchange. One technology, standard or not, leads to another:
some find a sweet spot at a particular time: no-one can tell exactly
whether a spec will succeed. If some specifications crash and die, no-one
should be surprised.
On the question of endless hours, in order to participate as a member
of a W3C committee your employer has to commit you to at least
one day a week fulltime and probably more, let alone teleconferences
(typically organized at times not suited for Eastern hemisphere) and
frequent meetings (typically held in the West). No comparison there,
W3C requirements are far more onerous.
As for difficult people, I found the ISO crowd without exception to
be polite and straightfoward, sometimes both. And, more importantly
to me, I never heard any description of people promoting "rival" approaches
being "paranoid", "conspiracy theorists" or their positions dismissed as "rants",
cowardly-seeming behaviour which I have seen repeatedly on W3C forums.
None of that is a criticism of W3C, or my many friends who are active
in its various efforts, b.t.w. Nor to dismiss Tim's experience.
What purpose does ISO bashing by a TAG member serve, apart from promoting
factionalism? Stockholm Syndrome perhaps :-) Tim calls a spade a spade,
but it is possible to have a souring experience anywhere, whenever one's particular
use-case is not recognised as being significant by a committee interested in taking
standards in a different direction.
When Tim promises that ISO people will be difficult, it reflects on the current individuals
involved: is James Clark difficult?, is Murata-san difficult?, is Jim Mason difficult?
is Komachi-san difficult? Difficult is not the first word anyone would use for
Ken Holman, who is a dedicated fount of gracious mildness. I guess the most passionate
advocate in the past has been Charles Goldfarb, but now that his ideas have won and are
secure he is avuncular. In any case, many of the old ISO people are now
involved in W3C, so presumably they would be being difficult in W3C too: but
when I read the contributions of Sharon Adler or Dave Peterson, for example,
to current W3C committees, they seem utterly constructive and pleasant
(as, indeed, Sharon and Dave are.)
To say that ISO committees may be difficult is to ignore that the participation in
any group based on position-advocacy and adversarial championing creates
difficulties for people from respect-based or seniority-based cultures. The almost
complete absense of active participation by non-Western Chinese in W3C, for example,
may be argued as evidence for that.
Finally, if the difficult one is me, let me apologise if I have let unpersuasiveness take
refuge in bad manners. But when and where?
> I'm assuming that those on the list extolling the virtues of SGML are
> already fully engaged in the process of keeping it alive, and thus
> speaking from experience. Right? -Tim
Well, I am, for what that is worth! And I also participate in W3C (for which I
thank Martin Duerst and the I18n WG.) XML and SGML only benefit from
each other, as far as I can see.
My company releases a markup editor next week. In talking to potential users,
SGML validation was commonly requested. SGML is still very strong at the
industrial publishing coalface: many people who have SGML document sets see little reason
to change (assuming that an XML system is not an XML system, which I don't
anayway. :-) XML certainly has taken over as the normalized form for automated
processing, replacing the hodgepodge before. And many people are sniffing around
XML adoption because content management systems may require it.
No-one I know is moving to XML Schemas: they need entities and therefore
DTDs anyway, so WXS does not give enough bang for the bucks.
We just had the Open Publish conference here last week, with mainly miltitary, legal,
heavy equipment publishers mixed with the advanced PDF crowd: it sometimes seems
that there is a greater commonality of interests between SGML, HTML and PDF people
compared to markup and database/messaging people.
Cheers
Rick Jelliffe
(Not writing on behalf of employer.)
[1] http://www.iso.ch/iso/en/aboutiso/introduction/index.html
|
