Re: Is DSSSL Syntax Tricky?

[Jacques.Deseyne@xxxxxxx (Jacques Deseyne)]

My apologies for this late reply -- I didn't read my mail earlier
this week.

Paul Prescod wrote:

>After thinking about it a little bit, I'm going to argue that the DSSSL
>syntax *is* a bit tricky (not out and out hard, but tricky) and is a
>disincentive for those who would make small-to-medium complexity DSSSL
>stylesheets without really learning the full DSSSL expression language.
>...

[Pertinent remarks on higher-order functions, lambda expressions, prefix 
 notation, difficult distinction between "data" and "code" ..., all
 behaving as a disincentive for newcomers]

I believe that for everyone with a minimum of Computer Science background
or equivalent interests, the expression language isn't so hard to learn, 
IMHO less hard than the proprietary languages coming with some conversion 
tools. Potentially, it is also much more powerful (and definitely more 
fun).

>This *does* slow you down in learning the language because it slows you
>in learning from other people's code. Also, I presume that I am not the
>only person who tries to look at a few examples of another language's
>code before I learn it, ...

Definitely not, but beware! There are examples of quite unreadable 
code in almost any programming language. 
Mastering a programming language is typically built up by an interaction 
between reading introductory materials, browsing through examples, 
doing hands-on exercises, reading books and the "official" language 
specification. Above all: using the language in a real project.

Analoguous remarks were heard during a tutorial on DSSSL Transformations 
at SGML Europe '97 last week, while a few simple examples were presented.
The main reaction from the delegates was "Why does it have to be so 
complicated while I can do this with my current 'xxx' or 'yyy' conversion 
tool using four lines of code ?"

Some elements for an answer are:

1) In really complex cases, a grove-based transformation could be more 
   easily specified (and verified) than a conversion using one of the 
   current tools (Anders Berglund gave the example of transforming 
   table markup).
   For Style specifications, DSSSL allows you to do much more than
   the average MS-Word template or other word processor style sheet.

2) Style or Transformation specifications could be generated by graphical
   tools, which can be "point-and-click"-driven or could use "simpler" 
   expressions as an input. Commercial tools should try to hide the
   language from the end-user (BTW, SGML tools may not have done 
   enough to hide markup from the end-user, either).

3) DSSSL is an ISO standard: it makes your application specification 
   vendor-independent (a rather theoretical argument at this time).


Surely many readers of this list must have better arguments ?


Best regards,

--------------------------------------------------------
Jacques Deseyne <Jacques.Deseyne@xxxxxxx>
SEMA Group Belgium - Stallestraat 96 - B-1180 Brussels

Tel + {32} 2 333 53 71
Fax + {32} 2 333 53 22



 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist