Re: draft-ietf-httpbis-digest-headers: Digests and Conditional Requests

[Julian Reschke <[e-mail]>]

On 08.07.2020 08:46, Mark Nottingham wrote:
>
>> On 30 Jun 2020, at 7:22 pm, Julian Reschke <[e-mail]> wrote:
>>
>> Hi there,
>>
>> I just noticed that RFC 4918 (WebDAV) and RFC 6920 (Naming Things with
>> Hashes) give us the machinery to make request conditional on hashes.
>>
>> Example (based on
>> <https://greenbytes.de/tech/webdav/draft-ietf-httpbis-digest-headers-02.html#server-returns-full-representation-data>)
>>
>>   PUT /x HTTP/1.1
>>   Host: foo
>>   If: <ni:///sha-256;X48E9qOokqqrvdts8nOJRJN3OWDUoyWxBf7kbu9DBPE=>
>>   Content-Type: application/json
>>
>>   {"hello": "world2"}
>>
>> All that would be needed is s statement that servers that support both
>> the Digest field and the If header field SHOULD consider the ni: URIs
>> built from the hashes as "state tokens" as defined in RFC 4918, Section 3.
>
> As I read it, that means you'll get a 412 when the hash matches. The opposite would be more useful (i.e., an analogy to If-None-Match with hashes) would be more useful for many cases...
> ...

Good point.

So make it:

   If: Not <ni:///sha-256;X48E9qOokqqrvdts8nOJRJN3OWDUoyWxBf7kbu9DBPE=>


> Having said that, I think a more specific header field would be more likely to get traction -- e.g., If-Not-Hash: ... That might enable several interesting use cases.
> ...

...already covered by "If". I would prefer using something that is
already defined, instead of having to mint two more header fields (we'd
need both "If-Hash" and "If-Not-Hash", right?).

Best regards, Julian