Lists Home |
Date Index |
Joshua Allen wrote:
> > "The creators of XML were wrong. Postel's Law has no exceptions."
A fine permathread. I often find myself defending Postel's law against
those who feel that they have discovered an exception.
Agreed. I find this a laughable sort of argument. An anecdote: I've placed
an order for the new dual G5 mac -- this will be the first desktop that I've
purchased since 1998 and will replace my dual P6 180 (Mhz!) running NT
4.0 -- now since I've got some time :-( before the machine arrives, I've
been busy reading "OS X for Unix geeks" and see that in Darwin one can
configure parts of the kernel/driver's etc via an XML config file (/me
breathes sigh of relief that XML is conservative in what it accepts). Did
this make me happy that XML is conservative in what it accepts ... I did't
give that a thought ... rather I thought "hey if there are XML parsers in
some OS kernel, XML really has become pervasive.
Now I'm also thinking, SMTP user agents that are so liberal in what they
accept are partly to blame for the rapid spread of email viruses (e.g.
buffer overflows etc.). Or otherwise, should password verifiers be liberal
in what they accept?
In any case back to thread. What is the point? Any of:
a) user agents should accept non-well formed strings that have angle
brackets and are 'close' to being XML.
b) we should forget XML and fall back on SGML which doesn't require eTags
c) user agents shouldn't require XML to be valid w.r.t a particular schema.
I'd say that Postel's law is easily applied to XML given the constraints of