Re: SAX 2.0 Enhancement proposal (2nd Ed)

[Paul Grosso <pgrosso@arbortext.com>]

I'm not on xml-dev--after many years, I recently decided I didn't
need the extra email in what has become a low signal-to-noise ratio
mailing list.

But, fyi, I am probably the most vehement supporter of the fact that 
SAX's entity resolver is wrong to (1) require an absolute URI and 
(2) return a stream as the result instead of just the resolved URI.

paul

At 14:36 2001 06 15 +0100, you wrote:
>To summarise the position so far:-
>
>There have been a number of eminent responders to my original proposal [1].
>Explicit supporters were Normal Walsh [2], James Clark [3] and Jonathan
>Borden [4].  Paul Grosso [5] did not say so explicitly, but I believe the
>proposal has his support also.  David Brownell [6] did not like the proposal
>at all, and David Cleary [7] is concerned about the effect that James
>Clark's modified proposal will have on non-Java implementations.
>
>The weight of opinion then is on the side of change.  The proposal for a new
>interface in the next version of SAX (SAX 3.0?) remains unmodified so I
>propose it goes onto the 'list of proposed enhancements' that I presume
>David Megginson has tucked away somewhere to be resurrected when SAX
>eventually goes to its new home.
>
>The proposal for a bug-fix to SAX 2.0 needs modification along the lines of
>James Clark's suggestion.
>
>Modified SAX 2.0 Bug-Fix Proposal:
>
>1) create a new standard feature:
>http://xml.org/sax/features/absolutize-system-identifiers with a default
>value of false.  Applications may modify this feature at any time.  When set
>to 'true', the behaviour of SAX will be altered so that calls to
>EntityResolver::resolveEntity will not turn system identifiers into absolute
>URIs.
>
>2) In order to furnish the application with the base URI of the entity
>containing the ENTITY declaration, I would like your opinion on a choice
>between:-
>
>a) When http://xml.org/sax/features/absolutize-system-identifiers is true,
>the systemId parameter contains the system identifier from the ENTITY
>declaration concatenated with the baseURI in effect when the ENTITY
>declaration was read.  The two strings are separated with SOH (0x01) which
>is not a legal system identifier character.
>
>b) A new standard property: http://xml.org/sax/properties/baseURI.  This
>property will contain the baseURI as described in (a) above.  The value of
>this property will only be valid during the invocation of a resolveEntity()
>call.
>
>Votes, suggestions and comments welcome!
>
>Regards
>Rob Lugt
>ElCel Technology
>
>
>[1]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00026.html
>[2]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00028.html
>[3]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00040.html
>[4]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00044.html
>[5]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00035.html
>[6]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00033.html
>[7]
>http://lists.oasis-open.org/archives/entity-resolution/200106/msg00045.html