W3C home > Mailing lists > Public > www-dom@w3.org > July to September 1998

Re: document.write to self during load not allowed by PR-DOM

From: David J Woolley <djw@bts.co.uk>
Date: Mon, 7 Sep 1998 18:07:58 +0100
Message-Id: <199809071705.SAA28168@saracen.bts.co.uk>
To: www-dom@w3.org
> methods.  If you are right then no one should be able to insert or delete
> nodes without invoking the open method in HTMLDocument.
> 

There is no precondition that "open" should be called for 
node.insert() that I can find, nor any obvious statement that there 
is an implicit call of document.open().  Although I've only searched 
the postscript (with the intelligent search tool in ghostscript) the 
only hits I found on open were for the document object and in the 
index.

If there is an implicit open, I think this should be explicit in the 
document.  I think it should also be made clear whether the initial 
load of the document can be treated as an implicit document.write, 
which would then tend to imply that document.write is fully 
recursive, i.e. document.write can create an immediate script 
fragment which itself calls document.write, directly or indirectly.
(Iterative may be more correct in that the rendering can be delayed 
and scripts can probably be deferred until rendering occurs.)

Note that Netscape's jsguide document says that document.write in a 
fully rendered document generates an implicit document.open which 
results in a new window, if an explicit one hasn't been performed.

An implicit open would still cause problems if insert/delete really 
required the document to be open in that sense, as the document need 
not be rendered until it is closed, whereas it seems to me tha the 
most useful time to use insert/delete is in response to events on a 
fully rendered document.

To some extent, though, the DOM is becoming academic as it implements 
a different model for modifying the document from that used by IE4 
(inner and outer HTML) and NS 4 seems not to have any equivalent 
facilities.  A the moment I can only use it to help define a lowest 
common denominator, where it is acceptable to our marketing people to 
avoid taking IE4 to the limit.  However, it was in trying to define 
this lowest common denominator that I hit the problem, as the W3C DOM
seemed to force a lowest common denominator without load time HTML 
generation.


-- 
David Woolley - Office: David Woolley <djw@bts.co.uk>
BTS             Home: <david@djwhome.demon.co.uk>
Wallington      TQ 2887 6421
England         51  21' 44" N,  00  09' 01" W (WGS 84)
Received on Monday, 7 September 1998 13:22:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 22 June 2012 06:13:45 GMT