I see that not(not()) now takes about 6000ms, which is definitely better, but still not what I would hope for. I also note that currently all the tests on https://www.wikifunctions.org/view/en/Z13499 timeout. But these are pretty arbitrary standards/expectations, so it's hard to say if there are still vast improvements possible.

Currently failing at https://www.wikifunctions.org/view/en/Z15091

But then it cleared up at about the same time that this happened: https://www.wikifunctions.org/wiki/Z15470?uselang=en&curid=28670&diff=98736&oldid=98731

Currently failing at https://www.wikifunctions.org/view/en/Z15473

T359233

Looking at this again, I wonder if this is ultimately because lists of natural numbers are not yet fully supported.

I don't understand why Z13528 (Test: 1+0=1) is on your list. That seems quite distantly related to this.

@ecarg should I move it back out of "verify in production"? Does it go to backlog?

It looks good in the summary mode:

Looks good in production.

Looks good.

Now I see the same error on a different test of the same function: https://www.wikifunctions.org/view/en/Z13334

Thanks for your work, it sounds like a good improvement. I look forward to testing it out.

Agreed. It passes now. I'll report again if I see a similar problem popping up again.

Yes, I think that is what the user means when they cancel publishing.

I found the semi-timeout in a simpler scenario, so am splitting that off as a separate bug: https://phabricator.wikimedia.org/T360385

IMO that does not solve the problem. What if you genuinely don't want to publish yet (so you want to cancel the publish), but you still want to work on your edited label and description a bit more before you publish? I think it should revert to exactly what you had before you pressed the "Continue to publish" button.

Here's some others which seem excessively slow.

• Go here: https://www.wikifunctions.org/view/en/Z10123
• Click the pencil next to "About"
• Change the "labels" field to "Much better Sandbox-Implementation 2 that took me 2 minutes to come up with"
• Click "Continue to publish"
• Click "Cancel"
• You're immediately back to square one, and have lost the cleverly crafted label you came up with.

I'm currently getting a similar issue at: https://www.wikifunctions.org/view/en/Z10006

<s>For me the test stays failed even in edit mode:</s>

I'm not sure what you mean by "scoping being mis-attached". But feel free to take this forward in whatever direction you think will help. I agree that Product/Design involvement is likely required.

Update: I was made aware of the in-built Reduce Function (https://www.wikifunctions.org/view/en/Z876), which enabled a hack to make apply2 (https://www.wikifunctions.org/view/en/Z13318). Unfortunately the Reduce Function is so slow that it runs out of time when checking if a three-element list is sorted (https://www.wikifunctions.org/view/en/Z13322).

I've added some tests to a few of those still without tests. It would help if someone could link them.

Maybe it's become deeper, but the initial issue seems to have been solved. I just created this composition: https://www.wikifunctions.org/view/en/Z13276

I can no longer reproduce this since the test case no longer calls the same function, and no longer fails. I don't mind if you close this, I'll report other evaluation failures as I see them (once I'm sure it's not my fault!)

Just adding a note that I've made a user-built function aiming for this same outcome: https://www.wikifunctions.org/view/en/Z13052

If you open the dropdowns you'll see that the function is selected. That interface display issue is reported as a different bug;

A similar thing shows up in the result validation section of this page: https://www.wikifunctions.org/wiki/Z11070?uselang=en&oldid=52133 where it says Enter String even though all fields are filled out once you open the dropdown.

So, this will affect any string function any time someone types in a QID as the input? I just tried entering "Q1" into https://www.wikifunctions.org/view/en/Z10012, and this seems to be the case. I guess I don't fully understand why it is "unsafe", but can't the software interpret it as a string if it's in a string variable field, but not if it's in a Z9 type field?

It would be good if the duplicity list showed the best-match WD label (for my language :)) and allowed a one click article-item merge. Many of the arz hits at the moment are galaxies with very distinct names, and even if they have slightly different hyphenation it would be easy to match many from the list without going through them individually.

Available in the Chrome Store: https://chrome.google.com/webstore/detail/entity-explosion/bbcffeclligkmfiocanodamdjclgejcn
On-wiki documentation and discussion: https://www.wikidata.org/wiki/Wikidata:Entity_Explosion
Open-source code: https://github.com/99of9/Entity-Explosion
Introduction video: https://www.youtube.com/watch?v=eRnqoJyi92w
Facebook page: https://www.facebook.com/EntityExplosion

Thanks for this. I like all your criteria, although I think I want an even tighter limit on #3, and am hoping for a little more pizzaz. For me "web" is a bit bland in that every extension is interacting with the www.

The name was rejected because it's "best to avoid using "Wikidata" (or other project names) as the first word in the app title". Has anyone got any other suggestions?

Eek, scary. Yes, here you go https://github.com/99of9/Wikidata-Connect . Please be gentle, it's still very rough, and I'm totally new to this game!

Here's a video demo of my first version.

Here's my similar first-step idea. You are browsing the web, and find a page on any other site about an item that should have a Wikidata item. You click a bookmarklet, which instantly tells you in a side panel which identifier property should be linked, and if it is linked, which Wikidata item the ID is linked from.

This used to work.

I don’t really like property_threshold ... with the output of « 0 (0%) » there is no way of telling whether this is an actual zero

I agree - I cannot imagine a situation where I would need this.

Looks good, thanks.

I've made progress on this. The page already calls a template ("Coloured cell") per cell, so my concern about template limits is irrelevant. So I decided to go with your "option 2", and made a helper template to generate these links. It has been set up as a wrapper for Coloured cell.

Or if that is too much, scrap the counts entirely (a percentage is already clear), and add two links with the symbols + and -. So it could end up simply as "86.56% (+ -)"

In the brackets you already show the number of items with this property. Hyperlink that. The second link for the negatives could also be in the brackets as a count, so it might look like "86.56% (2300:357)" where both the 2300 and the 357 are hyperlinks.

Does wikidata have a limit on how many templates can be on a page?

Thanks