W3C home > Mailing lists > Public > public-qt-comments@w3.org > February 2004

RE: [DM] IBM-DM-105: Order of comments, PI's and text given [schema normalized value] property

From: Michael Rys <mrys@microsoft.com>
Date: Wed, 18 Feb 2004 13:14:02 -0800
Message-ID: <EB0A327048144442AFB15FCE18DC96C70214D2AB@RED-MSG-31.redmond.corp.microsoft.com>
To: "Dimitre Novatchev" <dnovatchev@yahoo.com>, <public-qt-comments@w3.org>
Cc: "Michael Kay" <mhk@mhk.me.uk>

This was a very conscious decision during the design of the data model
and we had exactly these arguments. I personally would find it
problematic to change the semantics of the data model in this way to
help the <0.5% use case...

In general I agree with you about the need to preserve information. 

But the general way of typing an element node to be of a simple type and
allowing PIs and comments in the serialized form in my opinion was a
mistake of the XSD validation specification. Many people will utilize
such typing to map property-value pairs and could care less about
whether the original textual representation and an occasional comment is
preserved due to the fact that they operated on the typed data. If we at
least preserve the comment/PI, then the child axis access still gives
you the data...

The workaround is that you do not type your data and the comments, PIs
and text nodes stay where they are...

Best regards
Michael

> -----Original Message-----
> From: Dimitre Novatchev [mailto:dnovatchev@yahoo.com]
> Sent: Wednesday, February 18, 2004 1:08 PM
> To: public-qt-comments@w3.org
> Cc: Michael Rys; Michael Kay
> Subject: RE: [DM] IBM-DM-105: Order of comments, PI's and text given
> [schema normalized value] property
> 
> > If you provide data-centric application the ability to fold their
value
> > (and thus provide better performance) then this is one of the
> > consequences. And I think this is an acceptable trade-off...
> 
> Please, think and do not allow this.
> 
> It is harmful to allow producing incorrect results in the name of
"better
> performance".
> 
> In fact, the best speed for producing wrong results should be as close
to
> zero as possible. We should always do whatever is possible to decrease
the
> speed of producing wrong results.
> 
> Is it necessary to provide references to wise people, who said that
> without correct results speed is meaningless?
> 
> Best regards,
> 
> Dimitre Novatchev.
> 
> __________________________________
> Do you Yahoo!?
> Yahoo! Mail SpamGuard - Read only the mail you want.
> http://antispam.yahoo.com/tools
Received on Wednesday, 18 February 2004 16:14:15 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:45:18 UTC