ERB Decisions of March 26th

The ERB met Wed. March 26th.  All members were present except Dave Hollander,
who was represented by proxy [there had been *plenty* of advance discussion
of the items to be voted]; Peter Sharpe missed the final vote due to 
having to leave early.

1. Should the spec be changed to allow attribute values (specifically the 
   nonterminal QuotedCData) to include unsecaped "<"?

After some discussion, we were unable to develop any consensus in favor
of re-opening this question.  The spec stands as is.

2. XML requires the string ']]>' to be escaped as ']]&gt;' when it is
   data.  Should the draft specify that this is 'for compatibility' only?

Unanimous: Yes, this is for compatibility only.

3. Should the XML declaration at the front of the document entity be
   made optional?

[Ed. note: a *lot* of discussion on this one; made more difficult that
 the people who wanted it optional could see good reasons for making it
 compulsory, and those who wanted it compulsory could see good reasons
 for making it optional]

 Optional: Bray, Clark, DeRose, Kimber, Magliery, Sharpe, Bosak
 Required: Maler, Paoli, Sperberg-McQueen

 So it's now optional.

4a. Should we change the way the draft spec describes when and where
    parameter entity references are legal?  

 Unanimous: Yes.
   We have a proposal from Michael and me for major cleanups to describe 4 
   straightforward ways to use PE's, and much more controversial language
   for another more general way to use them.  The 4 straightforward ones
   are going into the [imminent] draft spec, the ERB still has to chew
   on the hard one.

5. Should production 69 (external ID) be changed to make the
   SystemLiteral optional?

 Unanimous: leave it required.

6. In section 4.2.2 "External Entities," should the following sentences
   be dropped (or modified)?

    barring an external mechanism for establishing the base...
    Relative URLs are relative to the location of the entity or file
    within which the entity declaration occurs.  Relative URLs in
    entity declarations within the internal DTD subset are thus
    relative to the location of the document; those in entity
    declarations in the external subset are relative to the location
    of the files containing the external subset.

  Unanimous: leave as is.

7a.  Should production 69 be changed to allow public identifiers?

No issue since DSD's has caused the ERB so much trouble.  The vote
went as follows:

 Yes, allow PUBLIC: Kimber, DeRose, Sperberg-McQueen, Maler, Hollander
  No, no PUBLIC ID: Paoli, Sharpe, Magliery, Clark, Bray, Bosak

So in this draft, no public IDs.  It should be voted that *every person*
on the No side would change their vote to Yes if there was an agreed-on
resolution mechanism for PUBLIC identifiers.

8. Should the predefined entities be removed or altered?

Proposal: Drop all predefined entities
Yes: Kimber, Bosak, Maler
 No: Bray, Clark, Paoli, Sperberg-McQueen, Hollander, Sharpe, Magliery,

Proposal: Well-formed XML docs are considered to have &lt;, &gt;, &apos;,
          &quot;, and &amp; predefined.  Valid XML docs must have them 
          declared *if* they use them; the spec will give a precise 
          definition of what the declaration must be.

Passed unanimously.  Outstanding item to get the declaration just right.

9.  Should we allow and ignore the tag omission [-O] [-O] syntax?  Another
    close call.  Pro: eases conversion and DTD management.  Con: 
    non-functional in XML, another irritant in explaining to the world.

Yes, allow them: DeRose, Kimber, Maler, Sperberg-McQueen
No, keep it as is: Bray, Bosak, Clark, Magliery, Paoli

So no tag omission indicators for now.

Cheers, Tim Bray
tbray@textuality.com http://www.textuality.com/ +1-604-708-9592