RE: Entity replacement

James K. Tauber (jtauber@jtauber.com)
Wed, 4 Jun 1997 23:22:39 +0800


> The % notation does, in effect, specify which non-terminal symbols
> can be replaced by p.e. references.

Not really because you get cases of %(...)

Now admittedly, productions that include this in their RHS could be
rewritten with an additional non-terminal symbol, so that production [43]
could be written

choice::='(' S? choicelist S? ')'
choicelist::=cps ('|' cps)+

And this is exactly what I would like to see done because you could then
simply list (apart from the productions themselves) those non-terminal
symbols that can be replaced by PEs.

Do other developers feel this would make it easier to go from spec to
implementation?

Now, relating my previous parsing/GE query to PEs:
Is it easy, given the current syntax spec, to build a correct parse tree of
a DTD before PE replacement?
If not, should it be?

James K. Tauber / jtauber@jtauber.com
Perth, Western Australia