cabo@cs.tu-berlin.de (Carsten Bormann) (06/27/91)
In article <JJC.91Jun27115235@jclark.UUCP> jjc@jclark.UUCP (James Clark) writes: I just started reading ISO/IEC DIS 10179, Document Style Semantics and Specification Language, and I was surprised to find that it made no use of the SGML LINK feature. The CD (DP) version of DSSSL did make use of LINK to bind a DSSSL specification to an SGML source document. The DIS version specifies in clause 2.2 "The binding of a DSSSL specification to an SGML source document instance is outside the scope of this international standard". While I did not take part in that decision, I could imagine that requiring support for a rather obscure% SGML feature such as LINK for the sole purpose of binding a style to a document was considered to be overkill (you could still use LINK for this purpose in the DIS version if you wanted). Now the next question is, "why doesn't DSSSL make use of LINK as a representation of association specifications?" (there actually was some pressure to use LINK as the basis for DSSSL in this way). The answer is that (a) LINK is rarely implemented in today's SGML products, and (b) LINK was not considered to be as powerful and easy to use as the location model (then called "FQGI") idea that underlies DSSSL. Consequently, DSSSL implementers would have had to newly implement an SGML feature that would not have been quite adequate to the problem anyway. A question whose answer would interest me is: who or what did make you aware of DSSSL, and how do you intend to use it? Gruesse, Carsten ----------------------------------------------------------------------------- % obscure in the sense of "rarely implemented" and thus not readily applicable to problems at hand.
enag@ifi.uio.no (Erik Naggum) (06/29/91)
Carsten Bormann relates that LINK was used for DSSSL in the beginning, but that the idea was rescinded after considerations such as: 1) it isn't much in use, 2) it isn't powerful enough, anyhow. The first is not argument. It's a simple fact (tragedy) of life that people don't do more than they need to, and since the objectives of LINK can be achieved partly with simpler means and less brain power spent, people go for the less demanding option unless they are forced to support something better. Therefore, we have such brilliant things like MS-DOS still among us. Had it been up to me, such staggeringly bad design should be outlawed, but we don't have enough courts to handle the load, even if they didn't have to mess with the sue-happy people, so it's probably best not to for purely practical reasons. If DSSSL had mandated LINK, LINK would have been implemented, and it would benefit us all. I use simple links a lot for various general output attributes, such as whether letters should be aimed for full- size envelopes, half-size or folded in three, which affects the placement and size of the address window. That said, I will respect the real argument in item 2, but I'd like to hear more about it. And what's FOSI? </Erik> -- Erik Naggum Professional Programmer +47-2-836-863 Naggum Software Electronic Text <erik@naggum.no> 0118 OSLO, NORWAY Computer Communications <enag@ifi.uio.no>