Re: ISSUE-4 (SpecContent): Should specifications decide what counts as content for transfer? [Progress Events]

* Anne van Kesteren wrote:
>Yeah, I'd very much prefer the Progress Events specification to handle  
>this so that not all other specifications using the Progress Events  
>specification need to do so. I agree that a protocol agnostic design would  
>be good, but that indeed doesn't preclude saying what should happen in the  
>HTTP case.

Then you should probably file a separate issue on this. It seems to me
though that saying much beyond that this is implementation-defined would
be misleading to authors, e.g., they might hardcode progress values in
their application and get very different values from the implementation,
breaking their application. This could happen e.g. if the user is behind
a proxy that rewrites the content in some way, or if there are errors in
the transmission (the browser may have attempted to pipeline several in-
dependent requests and has to retry them later due to a buggy server).

If not specified in elaborate detail, the specification should of course
highlight very explicitly that authors should not rely on the accuracy
of the numbers in any way.
Björn Höhrmann · ·
Weinh. Str. 22 · Telefon: +49(0)621/4309674 ·
68309 Mannheim · PGP Pub. KeyID: 0xA4357E78 · 

Received on Saturday, 21 June 2008 20:50:31 UTC