W3C home > Mailing lists > Public > w3c-sgml-wg@w3.org > May 1997

Re: XML FAQ: Answers needed

From: <lee@sq.com>
Date: Sun, 4 May 97 00:03:14 EDT
Message-Id: <9705040403.AA16768@sqrex.sq.com>
To: w3c-sgml-wg@w3.org
Peter Flynn <pflynn@curia.ucc.ie> wrote:
> a. Why did you change the format of comments to add the asterisks? I
>    have otherwise nearly-WF-HTML and this is annoying. 
>    [Didn't this get changed back? --PF]

If and when the suggested SGML TC is passed, SGML will support separate
start and end comment delimiters.  It will then be possible to use
{ and } (say) for comments, and to allow comments to include --.
I suggested the use of --* ... *-- so that -- could be allowed in
comments when SGML was changed, and until then, the XML comments would
work with existing systems.

Note that XML comments are not allowed in all places SGML allows them,
and even SGML does not allow comments everywhere.  XML is more regular
than SGML in this regard, as comments can be used everywhere you could
have a processing instruction.

The latest draft uses --....-- but I think this was an error that
crept in due to frantic editing.

> b. If XML does not use the RCS, but it may not include an SGML declaration,
>    how are non-XML SGML tools expected to handle the files? Is there a 
>    default XRCS?
Yes, there is an implicit SGML declaration.

> c. In the XML spec section 2.4 it specifies
>       [16] PCData ::=[^<&]*
>    Wouldn't the better way to state this is:
>       (CHAR)* -(<&)
>    Its confusing as stated.
Possibly.  Has this already been asked??

> d. An exemple of an XML DTD designed for a Web application is the Channel
>    Definition Format. [Didn't we decide this was invalid? --PF]
There is no such thing as valid XML yet, as the final specification
is not yet published.

> e. The SGMLDECL only calls out the . and - as LCNMCHAR and UCNMCHAR.
>    The Part 1 - Snytax defines:
>       [3] MiscName ::= '.' | '-' | '_' | CombiningChar | Ignorable | Extender
>    Shouldn't these two agree?
Yes, I think so...

> f. I could find no mention of the NAMESTRT variable in the Goldfarb SGML 
>    Handbook. What does this do as opposed to the LCNMSTRT and UCNMSTRT 
>    variables in the SGMLDECL?
There is no NAMESTRT variable in SGML, as far as I can tell.

> g. The statement that RCDATA, TEMP, IGNORE, or INCLUDE marked sections 
>    are disallowed in the instance is incorrect. The IGNORE and INCLUDE 
>    can be inferred via an entity reference. XML calls these conditional 
>    marked sections. [Is this right? --PF]
Originally in the discussion it was agreed to allow INCLUDEd and
IGNOREd marked sections only in the DTD; I think that features are
creeping in as the drafts are edited... :-(

For my part, I'd rather see us abandon marked sections altogether,
along with file entities, and use a preprocessor.

> h. The developers of XML should make a recommentdation to DTD developers 
>    to incorporate META data type tags at the beginning of their DTDs. It
>    would be of help to have a structure that would support retrieval 
>    and/or document management. [Are we doing metadata at any stage? --PF]
Perhaps an architectural form or a processing instruction is appropriate?
But this isn't a question!  I'm not sure what the bracketed comment means.
Probably XML should look at the work of the Dublin Core Metadata
workshops and the W3C Web Collections/dsig-manifests work.

> i. How does the application know the type of an attribute? The DTD has
>    to be present. So can a WF document use attribute types? Yes, if it 
>    contains an internal DTD subset, right? [Examples? --PF]
The instance can use any attributes; in a well-formed document, all
attributes have type "CDATA #IMPLIED", I think.  Except for attributes
whose name begins with -XML, which must have values beginning
with -XML and taken from a specific list.

Lee
Received on Sunday, 4 May 1997 00:03:22 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:26 UTC