Re: Progress reporting in future HTTP

George Phillips (phillips@cs.ubc.ca)
Thu, 11 Aug 1994 01:13:23 +0200

Reed Wade wrote:
>I _really_, _really_ like the idea of message codes; maybe I want
>a gas guage type progress indicator instead of a natural language
>message.

In theory, you don't need message codes to handle languages. The
browsers should be sending an "Accept-Language:" header and your
server should be using that to determine what to write in the
"Progress:" headers. And you don't need them in practice because
you either ignore the issue (98% of the cases) or your server has
already decided that the retrieval in question is part of the
English, French, German or whatever part of the document space.

I got bogged down with other things and never did come up with
patches to handle "Progress:" instead of "Status:" headers.
I contemplated having both a human-readable string form of the
header and a numeric "k of n" form. Certainly the readable
string form is more flexible so I recommend we just consider
that for now. "Progress:" is certainly useful when you need
it, but it's not so commonly necessary that it needs to be fancy.
Ideally it should never be necessary. The documents should come
back too fast.

Sure, it's a hack, but it's the simplest way to do it, is extremely
useful when you need it and beats the hell out of wondering if
the archie gateway is still processing your query or has, in fact,
been nuked from orbit. What needs to be done to get it in the spec?