Lists Home |
Date Index |
From: Tim Bray [mailto:firstname.lastname@example.org]
>On the contrary. View Source is essential. Read DTD is nice-to-have.
>And for something reasonably complex, like for example DocBook or HTML,
>in my experience it's *easier* to learn by View Source. Anyhow, I'd
>rather have a program read the DTD for me and tell me where I got the
>instance wrong. Reading moderately-complex DTDs with lots of parameter
>entities can be immensely difficult.
So can reading a few thousand instances to make a
statistical guess about "correct". Otherwise,
one accepts the browser's judgement and that as
we have seen, is a bit dicey.
In my experience, one wants both because one
can be trying to decide different issues of
correctness. What view source buys me is the
ability to rip off a construction that I can
verify in the browser of my choice is doing what
I want. Useful in the extreme but not definitive
given that my browser of choice may be doing
something funky, so I am back to view source,
a piece of ripped code and multiple browsers
I have to keep testing against. At least with
a DTD/Schema from the authority of record, I
can determine the difference between "correct"
and "working". With systems that will enable
both a system-bound definition and a mappable
public definition, this becomes even more
It's good to get the Web. It isn't good if the
Web gets me.
I agree about the parameter entities: a paper
chase like swing set assembly instructions.
The manufacturer says the parts are all there
but finding them in the bags is frustration
step by step, and God forbid one has to diassemble
it and move it.