On 03.02.2016 14:38, Daniel Kinzler wrote:
Am 03.02.2016 um 14:31 schrieb Markus Krötzsch:
Hi Moritz,
I must say that this is not very reassuring. So basically what we have in this datatype now is a "LaTeX-like" markup language that is only supported by one implementation that was created for MediaWiki, and partially by a LaTeX package that you created.
Markus, this TeX dialoect is not a new invention by Moritz. It's what the Math extension for MediaWiki has been using for over a decade now, and it's used on hundreds of thousands of pages on Wikipedia. All that we are doing now is making this same exact syntax available for property values on wikibase, using the same exact code for rendering it.
I think having consistent handling for math formulas between wikitext and wikibase is the right thing to do. Of course it would have been nice for MediaWiki to not invent it's own TeX dialect for this, but it's 10 years to late for that complaint now.
I do not agree with this argument. One could use a simplified version that is compatible with Wikipedia *and* with the rest of the world. We do not have MediaWiki markup in our text data, in spite of it being widely used on Wikipedia for many years -- instead, we now introduce a subset of it (the part you could put into <math>). If we have settled for a subset, why not use one that works with more commonly used tools as well? I don't think that MediaWiki LaTeX users would find it very hard to go back to the LaTeX they use elsewhere (in their own documents, on StackExchange, etc.).
A question you should ask when making extensions to the Wikidata data model is how much it will cost your data users to keep supporting Wikidata content in full. Such little twists, for a few extra commands, are creating extra work for many people.
The initial announcement one week before roll-out in the live system is not ideal either, adding some urgency to make this even more expensive. Now, four days later, even the final JSON datatype id for this has not been communicated yet ... we have 5 days left to update our code and make new releases. Of course, this schedule would leave no time for downstream users of our tools to update to the new version.
Data model updates are costly. Don't make them on a week's notice, without prior discussion, and without having any documentation ready to give to data users. It would also be good to announce breaking technical changes more prominently on wikidata-tech as well.
It would also be nice to include some motivation in your announcement (like "we expect at least 100K items -- 0.5% of current items -- to use properties of this type"). In the case of math, I could find some infoboxes that use LaTeX, so I guess this is what you are aiming for? I am not sure how many pages use such data though (inline math is of course very frequent, but it's not something you would store on Wikidata). If everybody can see why this is really needed, it will also increase acceptance in spite of some technical quirks.
Regards,
Markus
Moritz, I seem to recall that the new Math extension uses a standalone service for rendering TeX to PNG, SVG, or MathML. Can that service easily be used outside the context of MediaWiki?