Lists Home |
Date Index |
Gavin Thomas Nicol wrote:
> This is the point I was making. I heard from Paul, and others that we
> have to go mainstream, and use mainstream technology (ie. HTTP).
You heard wrong. The context was whether we should be worrying about
whether to use HTTP over TCP or UDP. Aiming for the mainstream would say
that TCP's latency is not a major problem for the sorts of things that
people want to do with web services. We could optimize right down to the
Ethernet layer in case someone wants to squeeze every bite of
performance out of web services but would that be a profitable way to
spend our limited time?
Those of us who advocate people understand HTTP do not do so because it
is mainstream. Rather because it demonstrably works and *already has* a
lot more design effort and thought poured into it than newer protocols.
Wouldn't it make sense to understand what you've got before you try to
> In XML there was "The Desperate Perl Hacker", in the DOM "The
> Desperate VB Programmer" and "The Hoards of Existing Users", in HTTP
> there was "Existing Practise"... all used to justify obviously, and
> admitted poor design. In most cases, these were phantoms and became
> largely irrelevant.
I don't think that's the case. First, XML never did a good job of being
accessible to the DPH so of course that guy never materialized. Whenever
I've tried to manipulate XML lexically I've run into huge problems with
e.g. CDATA sections, entities etc.
Third, backwards compatibility in HTTP is a totally different issue from
aiming for the mainstream. Apples and oranges.