Re: Resumable Uploads

[Martin Thomson <[e-mail]>]

This is certainly interesting work.  I find that HTTP uploads are too brittle.

One thing that I'd like to get a better understanding of is the
situation that lead to the need for resumption.

Is it the case that you have a PUT (or some other request) that was
interrupted in some way?  Establishing expectations on server behavior
with respect to that request is probably more important than
concentrating on the subsequent repair request.

--Martin

p.s., Content-Range and PUT don't appear to be the right choice,
though it might be possible if the server was able to indicate support
prior to the interruption.  A 1xx response might work; 100-continue
seems like a good place to indicate this, though it might be crazy to
tie the fate of this to something that is endangered as it is.

On 19 April 2013 04:42, Daniel Stenberg <[e-mail]> wrote:
> On Fri, 19 Apr 2013, Julian Reschke wrote:
>
>> "An origin server SHOULD reject any PUT request that contains a
>> Content-Range header field (Section 4.2 of [Part5]), since it might be
>> misinterpreted as partial content (or might be partial content that is being
>> mistakenly PUT as a full representation).
>
>
> This explanation basically rules out PUT completely for upload resume, as
> even if this would instead be done with an imaginary new header called
> Partial-update-of-remote-thing-please:, it could also become subject of
> getting handled as a full representation by mistake.
>
> And if PATCH is ruled out because how browser APIs, I guess only POST is
> left.
>
> (I personally don't think limitations in existing APIs are a very good
> arguments though, as we're surely forced to update things - including APIs -
> to take advantage of this once we agree on how things should work.)
>
>
>> If you believe that this is unreasonable, now and here are the right place
>> to discuss it.
>
>
> My personal belief used to be that Content-Range was a suitable header for
> this purpose, back with the original RFC2616 wording. With the updated
> httpbis wording it is clear that Content-Range doesn't work for this.
>
> --
>
>  / daniel.haxx.se
>