Looking at the code it is indeed impossible to pass an empty guid in
currently in SetClaim.php
The ChangeOpClaim class that SetClaim uses can actually handle no guid in a
claim and will create one for such a claim but it looks like we don't use
this in SetClaim, and can't.
The reason we can not accept an empty GUID is that wbsetclaim then has no
way of knowing where you would intend to create the claim.
If you take a look at the wbeditentity module you should be able to do
everything through there! i.e. pass in a claim with no guid and it will be
created.
Addshore
On 26 February 2014 19:01, Markus Krötzsch <mar...@semantic-mediawiki.org>wrote:
> On 26/02/14 18:41, Jeroen De Dauw wrote:
>
>> Hey,
>>
>> >> you can create claims with wbsetclaim. But you would need to create
>> a valid
>> >> GUID [1] yourself. The claim-GUID you send with your request needs to
>> be
>> >> <entityId>$<GUID> (e.g. Q2$5627445f-43cb-ed6d-3adb-760e66bd17ee).
>>
>> > Uh, didn't we fix this a long time ago? Client-Supplied GUIDs are evil
>> :(
>>
>> This has come up at some point, and as far as I recall, we dropped the
>> requirement to provide the GUID. So I suspect one can provide a claim
>> without a GUID, else something went wrong somewhere.
>>
>> > Requests with GUIDs that are not associated with any Statement should
>> be rejected
>>
>> Not allowing specifying GUIDs is another topic altogether. Disallowing
>> them is not the intention of the current system, and needs discussion
>> before that is changed. This seems to be both out of scope of this
>> thread and be way to low on the priority list (after all, where is the
>> problem, its been working just fine) to spend time on.
>>
>
> As Daniel said: "Client-Supplied GUIDs are evil :(". My only point was
> that one should never use such GUIDs for new statements. Returning an error
> seems better than just creating a new GUID, but both would work. Mainly, we
> should not use externally provided GUIDs. For example, if somebody is
> mistaken to believe that any string of the proper format can be used, then
> we might get bot code where the "GUID" is based on a constant. Wikibase
> should better make sure that it uses quality GUIDs, which are as random as
> a GUID is supposed to be.
>
>
> Markus
>
>
> _______________________________________________
> Wikidata-tech mailing list
> Wikidata-tech@lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/wikidata-tech
>
_______________________________________________
Wikidata-tech mailing list
Wikidata-tech@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-tech
