Just following up on some discussion I had with DanielK and Jeroen today on this, and summarizing it for the mailing list.

I still fail to see what the advantage would be to use the IRI datavalue - especially when it is basically stripped down to be a string datavalue, as Jeroen suggests in the last mail here.

I do see an advantage of stating the property datatype in a snak in the external JSON representation, and am trying to understand what prevents us from doing so. If we would do so, we would enable all use cases that were mentioned, or am I missing something?

I also recognize the importance of having soon an official JSON dump, the lack of which currently forces people to rely on the internal representation in the available dumps. As said, the format of the internal dumps is expected to keep changing. I will bump this up on the priority list accordingly.

(Re Commons media file being different: Commons media file is, in the end, also just a URI represented by a string, I do not see why it is so different to URIs).

2013/8/30 Jeroen De Dauw <jeroendedauw@gmail.com>
Hey Markus,

Thanks for the writeup. This clarified some things, at least for me.

However, this does not mean that you have to store the value as a compound object that contains many strings. In fact, this strikes me as a rather cumbersome approach that would make it harder to use the data. In SMW we store URIs as one string. Splitting this string into parts (under the assumption that it was a well-formed URL to start with) is quite easy, if this is needed (SMW does this). Conclusion: the use of a datatype for IRIs is in no way tied to the use of an impractical serialisation; reference implementations exist.

Agreed. The IriValue implementation is based on the SMW one, and retains this capability. Using serialize and unserialize will cause concaternation of the parts into one string, and then split them back up to a bunch before they are passed to the constructor.

So the solution to the problem at hand seems to be either to change these two methods to do the same as serialize and unserialize, or to simply not use these methods in our serialization process. The former approach is the most local and easy to implement, and given the urgency of this, the one I suggest going with.

(Somewhat different topic, mainly directed at the WD team itself:) It is however an indication that having these two methods in the DataValue implementations is not the best idea to begin with. This has been clear for some time, though in order to fix this, we effectively need to go with the second approach and implement proper serialization infrastructure for DataValues. That'd also fix a number of other problems and awkwardness the current approach is causing.


Jeroen De Dauw
Don't panic. Don't be evil. ~=[,,_,,]:3

Wikidata-tech mailing list

Project director Wikidata
Wikimedia Deutschland e.V. | Obentrautstr. 72 | 10963 Berlin
Tel. +49-30-219 158 26-0 | http://wikimedia.de

Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.