> On the whole I rather think, that the decission for eagerness was
> probably mostly carried by the arguments for simplicity and/or
> compatibility with the Scheme way of things, rather than by
> theoretical considerations for speed and equivalence...
I agree. Also, consider the simplicity of eager implementations compared
to lazy implemenations.
> Also I wonder if it would not still be possible (even if not really
> standard-conformant) to implement DSSSL in a lazy way, and probably
> most Style-Sheets would continue to work unchanged...
>
Yes, there is a theorem that proves that *all* stylesheets that currently
work would continue to work. Some which do not currently work would start
to work, too. In some sense it would be DSSSL++.
One other issue is that thanks to the eagerness, we can develop a
superset of DSSSL with side effects and know how it would behave. This
has already been proposed.
Paul Prescod
DSSSList info and archive: http://www.mulberrytech.com/dsssl/dssslist