Re: [Sks-devel] SKS, Content-Length and HEAD requests

[David Shaw]

On Nov 2, 2010, at 5:45 PM, Jonathan Oxer wrote:

> On Tue, 2010-11-02 at 13:20 -0400, Phil Pennock wrote:
>> On 2010-11-02 at 16:20 +0100, Marian Kechlibar wrote:
>>> The main principle would be to issue a HEAD request first, instead of
>>> a GET request.
> 
>> You've just added an extra round trip for the cases where the request is
>> specific, which makes things slower.
> 
> My reading of Marian's suggestion was that it would be optional, not
> required, so clients that are operating in constrained conditions could
> choose to do the HEAD request followed (depending on the result) by a
> GET request. Other clients could operate as normal by issuing a GET
> request initially.
> 
> Phil's suggestion of a limit variable seems good, but there's also a
> third option that may be worth considering: a command specifically to
> retrieve metadata about a query result without retrieving the result
> itself. There could be a command issued as a GET request in the same
> format as a regular search, but rather than returning the result it
> would return things like the number of results and the content length of
> the full result set in a machine-parsable format. Or it could just be an
> argument appended to the regular search.

I like this idea.  I think you could implement this in terms of Phil's limit 
idea, too: just request a limit of 0.  Since Phil's design always responds with 
a X-header containing how many hits there were (regardless of how many were 
actually returned), then you get the information you were looking for.

David