Matthew Cruickshank wrote:
>
> >IOW, if you design a browser to accomodate a 20% error rate,
> >designers will allow pages to pass with 25% bad code. If you
> >change to 25%, the designers will allow 30%. It's a negative
> >feedback loop.
>
> I can't think of any reason why not closing 100 tags would be of
> benefit to anyone (well, aside from bandwidth trolls!). Of course an
> author should fix that. But as a browser I don't believe it should
> have any limit to the types of errors it should attempt to overcome in
> order to finally render a page - certainly not something as trivial as
> a fixed number of unclosed tags.
>
> For me, personally, I wouldn't recommend any software that can't cope
> with broken input - that's a flaw, not a political statement.
That is an interestingly broad statement. Do I understand that there are
no high level language compilers or interpreters that you can recommend?
Does this equally indicate that you cannot recommend any assemblers and
macro generators for MCU's and MPU's? It is a very good thing that you
do not program computers. Assemblers, compilers and interpreters tend to
be persnickety. I once asked a compiler designer why the DEC compilers
were smart enough to isolate errors but never corrected them rather
throwing numerous errors. He said they had experimented and found that
they could not anticipate all possible errors. The idea is that if you
idiot proof something, there comes a better idiot.
Chuck
--
... The times have been,
That, when the brains were out,
the man would die. ... Macbeth
Chuck Simmons [EMAIL PROTECTED]
