Re: API Rate Limits and HTTP Code [#255]

That's the concern underlying my question; scoping metadata in HTTP headers -- especially existing ones -- is often an exercise in "I meant what I said, and I said what I meant." 

While it's fine for Bryce to document what he means for interested developers to figure out, it's less possible for automated, generic software (e.g., a robot, spider, cache) to understand what concepts of user identity and metadata scope he has in mind. And, that's the point of defining it as a standard. 

Furthermore, if we define a scope that's sufficiently precise to be good for Bryce, it's very likely that it won't be good for someone else who's minted a different API.

At this point I'd observe that HTTP doesn't disallow Retry-After on a 2xx response; only that it doesn't define what it means. As such, Bryce is free to define additional semantics for it (or if that is distasteful, another header) so that people who read his documentation will know what to do with it. However, as per above I don't see how we can explicitly allow it without falling into this trap (beyond perhaps having some general introductory text about what is and isn't permissible in cases like this).

Those are my thoughts at the moment; it's late here, so I'll stop there.


On 30/01/2011, at 12:41 AM, Karl Dubost wrote:

> 
> Le 29 janv. 2011 à 00:04, Mark Nottingham a écrit :
>> It seems like you want the scope of the 503 on the Retry-After to be the entire server, correct? I.e., on a 200, its semantic is roughly "here's your answer, but don't ask me (the server) another question for n seconds."
> 
> 
> Given the example below I would say for a group of URIs pattern more than the full server. you could set for sure api.example.com but if the server is done with a 
> example.com/api
> example.com/profile
> or even
> example.com/api2
> 
> And there are different limits based on your subscription levels: Premium for api and silver for api2.
> 
> 
> Le 28 janv. 2011 à 14:19, Bryce Nesbitt a écrit :
>> I'm talking about incremental retrieval of a large data set.  Each
>> resource is new to the client, and will not be cacheable:
>> 
>> GET /oai2.cgi?token=Az123Z&offset=0&count=100&query='Obama'
>> HTTP/1.1 200
>> Retry-After: 10
>> Cache-Control: no-store, no-cache, private
>> 
>> GET /oai2.cgi?token=Az123Z&offset=100&count=100
>> HTTP/1.1 503 Excessive queries from this IP address.  Please respect
>> the Retry-After header.
>> Retry-After: 10
>> Cache-Control: no-store, no-cache, private
> 
> 
> 
> 
> 
> -- 
> Karl Dubost - http://dev.opera.com/
> Developer Relations & Tools, Opera Software
> 

--
Mark Nottingham   http://www.mnot.net/

Received on Saturday, 29 January 2011 13:51:04 UTC