[Developers] minor change in search api behavior (empty result set will no longer be an error)

[Andi Vajda]

On Thu, 17 Jul 2008, David Flanagan wrote:

> I've always been told that the correct way to test for success with
> Metaweb services is not to test for a code equal to /api/status/ok, but
> to check for a code that begins with "/api/status/ok".
>
> It occurs to me that this is a case where it might be useful to return a
> code /api/status/ok/empty.  It still indicates success, but allows it to
> be special-cased, when desired.

+1 !

Andi..

>
> (And those of us who write documentation saying "you have to do a prefix
> match on the code" can give an example that proves the point!)
>
> 	David
>
> brendan wrote:
>> Hi all,
>>
>> This change is no likely to affect any apps/scripts out there, but here
>> goes:
>>
>> Currently, if you make a search request and the service finds nothing,
>> it comes back as an error.  e.g.:
>>
>> http://freebase.com/api/service/search?prefix=insx&type=/music/artist&type_strict=all
>> <http://freebase.com/api/service/search?prefix=insx&type=/music/artist&type_strict=all>
>>
>> {
>> "status": "200 OK",
>> "code": "/api/status/error",
>> "messages": [
>>  {
>>    "message": "No search results found",
>>    "code": "/api/status/error/search/empty"
>>  }
>> ],
>> "result": []
>> }
>>
>> In the coming weeks this will change.  The expected response for an
>> empty result will be:
>>
>> {
>>   "status": "200 OK",
>>   "code": "/api/status/ok",
>>   "messages": [
>>     {
>>       "message": "No search results found"
>>     }
>>   ],
>>   "result": []
>> }
>>
>> This response just seems more sensible, particularly given that the
>> service allows the user to request a very constrained result (strict=all
>> means results must match the given type parameter value)
>>
>> If your code has error handling for this "no results" case, you might
>> want to make sure this change won't break things.
>>
>> Let me know if you have any questions/concerns.
>>
>> Brendan
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Developers mailing list
>> Developers at freebase.com
>> http://lists.freebase.com/mailman/listinfo/developers
>
> _______________________________________________
> Developers mailing list
> Developers at freebase.com
> http://lists.freebase.com/mailman/listinfo/developers
>