Re: Draft for Resumable Uploads

>

 >>
>> A resource has to exist first, before it can support PATCH. 
>>
 >
> Says who?
>


Common sense? Clearly-defined method semantics is part-and-parcel of a uniform interface. If we're going to muddy the waters by allowing partial PUT (or PUT no content to DELETE), and PATCH to create a primary resource (not sayin' PATCH can't result in a /previous-version resource being minted), then I guess HTML was right all along to only bother defining GET and POST in forms.



>

> Some origin servers support use of the Content-Range header field

> (Section 14.4) as a request modifier to perform a partial PUT, as

> described in Section 14.5.

>



Ugh. There's some wording above that, about how a service SHOULD use POST instead of PUT; I'd suggest (sometime in the future if PATCH is ever officially added) similar wording about SHOULD use PATCH. Some origin servers treat a zero-payload PUT as a removal request; similarly, some wording about SHOULD use DELETE could clarify.



I managed to watch a lot of golf this weekend while reviewing the latest HTTP drafts. Aside from the above, the only other editorial suggestion I have would be to add "8.4.1.4 Brotli Coding".



-Eric