Lists Home |
Date Index |
> If anything, the stated rationale gives me hope that either:
> a) there are some misunderstandings about what is actually required of
> a non-validating processor (as Elliotte speculated), or
> b) the real reason for the decision is not so much footprint
> restrictions as it is the reason actually stated in the spec (in which
> case further scrutiny may be of help).
>  <http://jcp.org/aboutJava/communityprocess/review/jsr172/index.html>
It seems to me that what they wanted to achieve could be done by requiring
that all XML documents for J2ME be labeled as standalone, and if they were
not, the parser should notify the application so that the app could decide
whether to proceed or not. Then they could call the parser a
"XML-standalone" parser if they wanted to, meaning a parser that followed
the processing rules in the XML Rec for standalone DTDs. It would be easy
to understand what would or would not be processed, and nothing would
automatically break just because harmless DTDs were received.
It probably would not increase the footprint much for the parser to try to
process a DTD that was not labeled as standalone, and only notify the app if
it actually encountered some non-standalone construct. I think that the
majority of DTDs are standalone but not labeled as such.
This approach would not break when harmless documents with DTDs were
received, and would not claim to be XM-conformant when it was not