Re: lynx-dev OBJECT_TYPE (was Re: http://www.quantum.com/)

[Klaus Weide]

On Sun, 18 Jul 1999, Michael Warner wrote:
> On or about 16 Jul, 1999, Klaus Weide
> <address@hidden> wrote:
> > I see a completely empty page when I load http://www.quantum.com/.
> > But that looks lik Lynx's fault.  The contents start with
> > 
> >    <OBJECT_MAP></OBJECT_MAP>
> > 
> > which makes lynx ignore everything after it.  According to TRACE it's
> > looking for a closing </OBJECT>, not an </OBJECT_MAP>.

> So does lynx's error lie in not looking for </OBJECT_MAP>, or in
> not ignoring the pair entirely?

The basic error is that lynx has a wrong understanding of what characters
can make up element names.  It accepts only (US-ASCII) letters and numbers,
so it regards the '_' as ending the name.  So lynx thinks it got an
<OBJECT [+some garbage]>.   Lynx should also accept ".-_:" as name
characters.

The problem is then compounded because lynx parses OBJECT content in a
special "litteral" (sic) way (not according to any HTML standard afaik),
which means it is looking exactly for the string "</OBJECT>", and it also
seems to throw everything away if it cannot find that before the end of
the text.

What lynx should do is of course ignore this completely, as other
unrecognized tags.

> What *is* an OBJECT_MAP, anyway?  

I don't know, and don't care much...

> Does it exist in some DTD, or
> is it some sort of private extention?

Those don't exclude each other, since everyone can make up their
own DTD.  Make a DTD that has it, and it will exist in some DTD...

   Klaus