RE: Use of 1*DIGIT

Roy T. Fielding said:
> Implementions need to handle large numeric strings no matter how
> large they might be, regardless of what the protocol says they
> should be, and generally do so by returning an error if the
> number is larger than the maximum for the internal representation
> used for the value.

That's fine and dandy at the API level. I'm asking how that error should be
represented at the protocol level. Should the client close the stream?
Should the client try again? Should the server send a 5xx? Or a 4xx, since
the request needs to be modified? Should the server close the stream?

> This will change over time (as data get bigger)
> and may be much larger for specialized implementations than it
> would be for general-purpose implementations.

Regardless, the two should be able to interoperate, since they're
purportedly both speaking HTTP/1.1 -- but that won't happen unless the spec
is implemented to the T, using a representation that can hold an arbitrarily
large value (which I'd argue that most implementations won't do, because in
general it is an overly-burdensome requirement).

> 10 years ago almost
> everyone thought that 4GB would be a reasonable limit for an
> implementation of Content-Length -- now that is clearly not the
> case for the video-on-demand folks.

Yeah. I remember wondering how I was _ever_ going to fill that shiny new
100MB hard drive.


-- Travis

Received on Friday, 22 December 2006 00:56:54 UTC