Wikidata

Wikidata:Contact the development team/Archive/2019/01

< Wikidata:Contact the development team‎ | Archive‎ | 2019
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

2019 development plan (bis)

I like the new presentation of the development over the next year, I just hope that the plan can be followed and that updates during the year will show the real progress of the different tasks.

Only one comment about the signed statement proposal. From my point of view this proposal mix two different concepts: the signed statement and the link between the value and the reference. If the signed statement is good from data donation point of view, I find that the link between a value and the reference in a statement has to be available to each contributor and not only for some institutions. There are many contributors working hard to provide good references to the values they add and allowing other contributors to modify the value or the references without any indication of the change is really painful. I really support the proposal, but I think that offering some advantages to some institutions only and not to contributors is really a lack of respect to those who are curating the data and trying to maintain some high quality level in their work.

Tracked in Phabricator
Task T76232

I just mention a phabricator ticket pointing that problem and I would propose, instead of a signed statement for institutions only, to implement a locked statement concept: when a value is added with a reference, a protection is added to prevent the modification of any part of the statement. The locked statement can only be deleted. If a part of the stament has to be modified, then the statement has to be deleted completely and then created again with the correction included.

IMHO, the Wikidata model was badly built: a value can't be set up without a source (only instance/subclass statements should be created without reference as these two properties are used for classification and relation between items). Allowing the creation of a value without reference or allowing the modification of the value or the reference only was a bad idea and is the source of most of our quality problems. So instead of creating this signed statement as a patch, perhaps a deeper solution should be studied. I am not a specialist, but I think that a tag in the statement and a modification of the editing functions to handle that tag according to the above process is sufficient to implement the locked statement concept. This can improve dramatically the data quality in WD and avoid different edition abilities between contributors. Snipre (talk) 20:00, 28 December 2018 (UTC)

Hey Snipre :) I don't doubt that there are a considerably number of editors that add valuable references and <3 for every single one of them. So apologies if it came across as disrespectful. It wasn't meant that way at all. Signed statements as I imagine it is at first targeted at institutions but there is nothing that would prevent the community from setting things up in a way that also individuals that are trusted can sign statements. We'll have to hash out a process for this but I don't imagine anything built-in that would prevent it. I do currently imagine it to be useful especially for mass-uploads (though more can happen later) but that can be done by anyone really. I do focus on institutions because I want a signature also to be a signal to the outside. They should be able to see that an institution they trust has contributed data and has said that the data you see is what they provided. This is a stronger message to the outside than one of us going and saying "I checked that this data we have is the same as the data out there". Both are valuable but in different ways. --Lydia Pintscher (WMDE) (talk) 15:10, 30 December 2018 (UTC)
@Lydia Pintscher (WMDE): Thank you for your answer. I have to admit that my words are strong ones but your proposal is really a small step in regards of the important demand for high quality data in WD. Again my problem is not the signed statement but the logic behind it: you assume that a data can be considered as good data because the contributor who imported tehm can be trusted. This is the inverse of the wikipedia spirit where the information can be trusted not because of the authority of the contributor but because of the possibility to check the origin of the information by accessing the source through the reference description.
With your proposal you have no insurance that the data quality will increase: if the institutions don't provide the full description of the value metadata (qualifiers, publication date, versioning, ...) their signature won't increase the trust in WD. If no reference or incomplete reference are provided, there will be no way to check later the source of the data, this can be a problem if there is an error in the data import, if the data is changing over time or if two institutions are providing contradictory values.
Your proposal is pointing a fondamental point in data quality: the link between the value and the reference. But your proposal is not generalizing this link and that's all the weakness of that proposal.
Again, the idea of the locked statement in from my point of view a better solution than the proposal by offering the insurance that the imported data can't be manipulated. By providing a good reference, the institutions have by the same way the insurance that they will be correctly credited: a reference which can't be modified can act as signature. And the reference offers an advantage compared to a signature: a reference provides you the possibility to check the correctness of a value by comparing the value displayed in WD with the original value. A signature can't provide that possibility. Snipre (talk) 21:21, 2 January 2019 (UTC)

Image can't be find with exact title

Tracked in Phabricator
Task T196165

Hello, I have reported this bug a while ago and I thought it was corrected, but I have it again. When trying to add File:Route entre Barcelonnette et le col de la Cayolle (D902) -7.JPG to Église Saint-Laurent de Fours (Q38656180), I select image (P18), paste "Route entre Barcelonnette et le col de la Cayolle (D902) -7.JPG" but can't find the right image in the list... Ayack (talk) 10:55, 8 January 2019 (UTC)

Hello Ayack, and thanks for pinging us again about this issue. We included it in our workflow for the incoming weeks. Lea Lacroix (WMDE) (talk) 08:56, 10 January 2019 (UTC)
Merci Léa ! Ayack (talk) 09:38, 10 January 2019 (UTC)

Undoing a revision leads to dupplicate of values for a statement

I can't find a bug tracker so please let me know if it is not the right place to put this issue. I undo an 'instance of' value of Q845774. It replaces this 'instance of' by the one existing at this moment (Q194188), which leads to a dupplicate since this item already had this value for this statement. 1. I think it is a wrong behaviour, the Q194188 should not have been added since it was already there. If you agree, can a fix be made? 2. I can delete one or the other but which one preferably in order to save history? Thanks for your support.

Hello,
I general, the software is not preventing the user to add two times the same value in a statement. The fact that you can, by undoing an edit, arrive to the result where two values are identical in the statement, is also not prevented by the software.
Did I correctly understand your question?
Are you suggesting that the interface should not allow users to enter two times the same value in a statement?
Thanks, Lea Lacroix (WMDE) (talk) 13:38, 10 January 2019 (UTC)
Thanks for the feedback. I've nothing against the fact that a value can be set twice (I did not think about any implication at the moment) but it leads to the point where I'm not sure which one to delete (one is linked to another member edit, one is linked to my undo). I deleted the last one of the list thinking that if there is no clue then it is safe to delete whichever we want. Yes the question is answered.

Interface administrator right on test.wikidata

Hey. Could a bureaucrat on test.wikidata.org give me the interface administrator right overthere? I would like to test out some fixes related to Wikidata:Tours. --Pasleim (talk) 14:35, 14 January 2019 (UTC)

@Pasleim: Should be done now :) Let me know if you encounter issues. Lea Lacroix (WMDE) (talk) 10:46, 15 January 2019 (UTC)
Thanks. --Pasleim (talk) 10:49, 15 January 2019 (UTC)

Remove all redirects from watchlist

Hi! I'm not sure it's the right place, excuse me if it's not. I have a question about Watchlist: is there any way to remove all the redirects from the Watchlist in one click? If not, would it be possible creating it? I have more than 50k pages in my Watchlist, so I'm not able to open Special:EditWatchlist in order to manually remove them, and I don't know other ways. Thank you very much, --Epìdosis 11:21, 16 January 2019 (UTC)

Hey @Epìdosis:, I don't have the answer to this question, so I'm going to investigate and hopefully come back to you soon :) Lea Lacroix (WMDE) (talk) 10:05, 17 January 2019 (UTC)
Hello again!
So: there is no easy feature that would allow you to remove all the redirects from the watchlist. I've been looking for some alternatives you can use.
  • you can add a small css hack to your common.js, highlightling the redirects, which would help you removing them progressively when they appear in your watchlist
  • you can access the raw list of you watchlist, save it into a file, an then run a script that would query the mediawiki-API to check if each page is a redirect, and remove all of them
  • you can look at the code from the Pywikibot watchlist feature that could help you achieve that
I hope that helps a bit! Lea Lacroix (WMDE) (talk) 13:21, 17 January 2019 (UTC)

@Epìdosis: Open PetScan. Copy your raw watchlist→Other sources→insert to manual list→set wiki as Wikidata→Page properties→all namespaces→set redirects as no→Categories→choose Wikidata→Do it! The result will be your watchlist without redirects. Bencemac (talk) 09:51, 19 January 2019 (UTC)

@Bencemac: Thank you, it worked perfectly! And thank you also @Lea Lacroix (WMDE):. Bye, --Epìdosis 10:09, 19 January 2019 (UTC)
Thanks Bencemac for the efficient tip! :) Lea Lacroix (WMDE) (talk) 09:07, 21 January 2019 (UTC)

Request

Hello.Please use 

{{int:Thank-you}}

instead of {{Thankyou-tag}} in Template:Thank you and c:Template:Thankyou.Thanks David (talk) 08:11, 21 January 2019 (UTC)

Hello! Can you explain a bit more why this change should happen?
The development team usually doesn't change the templates, it's something the community takes care of. Maybe someone wants to help with that? Lea Lacroix (WMDE) (talk) 09:10, 21 January 2019 (UTC)
I want to dispense {{Thankyou-tag}} and this edit needs technical expertise David (talk) 11:57, 21 January 2019 (UTC)
  Done --Pasleim (talk) 12:49, 21 January 2019 (UTC)

Small modifications

Hello, thank you at first for this very useful wikidata! I'd ask some functionalities, don't know whether their are difficult to implement or not :

  • Order declaration per annum (see Zürich Airport (Q15114) on its patronage (P3872) or present them in a looking-like table, more easy to read, sort out and verify
  • Make clearer level of rank for a statement (preferred = make values bold, normal, normal and deprecated italic and smaller font for instance)
  • Make a +/- toggle button (reduce/open) for very lengthy statements (exemple France (Q142) is a very long page, it would be nice that by default every statement is reduced)

Thank you again for the work! Bouzinac (talk) 12:17, 21 January 2019 (UTC)

Hello Bouzinac, and thanks for your suggestions.
Over time, the community has build gadgets and scripts to enhance the interface and make it more efficient for them. You can look in the gadget list and in Wikidata:Tools to see if you're looking for is already covered. Note that all the tools are not necessarily listed in there, and people can also have customized scripts on their accounts: if you don't find what you're looking for, feel free to ask on Wikidata:Project Chat. Lea Lacroix (WMDE) (talk) 14:15, 21 January 2019 (UTC)

Edit filter to track and possibly disable changes to referenced statements

Can you add something to for this type of edit? It changes a satement supported by 5 references. --- Jura 09:39, 28 December 2018 (UTC)

Hello, the abuse filters are defined by community members. I ping @Matěj Suchánek: who made a lot of them. Lea Lacroix (WMDE) (talk)
This is an endless loop. I was the one who pointed out this hasn't been possible with abuse filters... Matěj Suchánek (talk) 09:08, 30 December 2018 (UTC)
Thanks for your answers. We'll follow up on that in the next weeks. Lea Lacroix (WMDE) (talk) 12:50, 30 December 2018 (UTC)
I think we need to define what we want here. Do you want an edit tag? Do you want a pop-up in the UI that nudges the editor? Something completely different? --Lydia Pintscher (WMDE) (talk) 14:59, 30 December 2018 (UTC)
My preference would be disabling such an edit but this could become another subject to glitches, misunderstandings or so. Perhaps a pop-up with a confirmation? At least for newbies. An edit tag could also work. Matěj Suchánek (talk) 16:05, 30 December 2018 (UTC)
  • It should be made possible to access these edits in the edit filters. There it can be decided what to do. In general, one can probably still add qualifiers, but shouldn't change the main value. --- Jura 08:20, 31 December 2018 (UTC)

Get Person Data from URL

Dear Ladies and Gentlemen,

I would like to read the data by URL call. As in this example: http://www.wikidata.org/wiki/Special:EntityData/Q76749 However, I do not get all the data here, which can be seen on the page https://www.wikidata.org/wiki/Q76749.

Do you have a tip on how I get all the data or what am I doing wrong?

Yours sincerely Christian Schönefeld

The URL does content negotiation and gives you the most appropriate format. If you want to force a specific format simply add it to the end of the URL. Example: http://www.wikidata.org/wiki/Special:EntityData/Q76749.ttl or http://www.wikidata.org/wiki/Special:EntityData/Q76749.json --Lydia Pintscher (WMDE) (talk) 14:50, 29 January 2019 (UTC)

P996 for every digital text

The current label and description of document file on Wikimedia Commons (P996) show that it should only be used for scanned texts, but there is no property for texts that are not scanned. I proposed an Item for non scanned texts. But the discussion says it would be better to rename the property for scanned files. --GPSLeo (talk) 19:36, 28 January 2019 (UTC)

That's something you best discuss with the editors either on Wikidata:Project chat or on the talk page of the property. It's not something for the development team to decide and do. --Lydia Pintscher (WMDE) (talk) 14:54, 29 January 2019 (UTC)

Datatype to external-id

Please, change datatypes to external-id. -- VlSergey (трёп) 11:24, 25 January 2019 (UTC)

Hey :) Happy to. Is there some sort of consensus to do it? --Lydia Pintscher (WMDE) (talk) 12:38, 27 January 2019 (UTC)

Daty Wikidata Editor, alpha release

Hi everyone,

I am Pellegrino Prevete, aka Ogoorcs and I am proud to officially announce the alpha version (Q2122918) release of Daty (Daty (Q60949478)), the native Wikidata editor I proposed at the Ideathon of itWikiCon 2018 (Q43527331), which aims to hugely simplify Wikidata UX for new and old advanced users.

During this first development month, as hoped, Daty has found approvals outside of wiki communities, too: the GNOME (Q44316) project has in fact accepted to host it on its development platform and the software has already been published on Flathub (Q43089335), the free software GNU/Linux app store in Flatpak (Q22661286) format.

Unfortunately I was not able to pack all planned features in this first release, although I hope that, trying it, you will agree that the work done has been adequate.

Set up sound foundations for the program was where it took longer than expected, i.e. make it work on all supported platforms and on all screen format factors. In fact at the time of writing Daty is one of the few responsive GTK (Q189464) applications and the only cross-platform one.

To calm down the potential storm of people fearing for vandalisms caused by a simpler editor, I must warn you that until an adequate revert tool for mass edits made with the program will be made available, Daty will browse the database *read-only*. At this time already it has been made so (not specifically in Daty) that only registered users will be able to edit entities.

Download

Installer links are available for Microsoft Windows (64 bit) and GNU/Linux (all architectures).

You can read a more complete changelog on my blog; bug reports can be sent on the issues page.

Note for GNU/Linux users

If you use a Flathub-integrating distribution (Linux Mint, Endless OS and others), you can directly install the software from your graphical package manager. If your distribution preinstalls GNOME and GNOME Software (Q15968880), you will just need to open the *Activities* screen and search for "Daty", as seen in this picture.

In any case you can install flatpak on your distribution by visiting this page or follow the distro specific installation istructions on the Daty homepage.

If you already installed a previous flatpak of the software, I advice you to wait for the update of tomorrow (build already scheduled), because of a last-minute bug in the configuration directory permission settings which has been corrected this morning.

Note for Ubuntu users

Since at this time Ubuntu has decided to support by default only the Snap (Q22908866) package format, you will not directly find the program in the software center. If there are enough requests though, I will make a snap version of Daty.

In any case deb (Q305976) packages will be made available in due time.

Note for Mac users

The software works on Mac, but since I do not own one I could not create the executable file. Again, if there are enough requests, we can find a way to solve this.

Thanks

First of all I want to thank Wikimedia CH for trusting the idea; without them Daty would still be a mockup this day. I hope that the global community, as the Italian one already did at the ItWikiCon Ideathon, will see the impact and the usefulness of a native editor, to please advanced users and greet new ones.

Of course I have to thank the GNOME project, which accepted the project on its infrastructure, and its developers, volunteers and contributors, who saved me from many headaches this month and before. I think it is a really great community.

Ogoorcs (talk) 01:32, 30 January 2019 (UTC)

Monolingual string language code ("cnr")

Please add "cnr" for Q7700307#P1705. I found it on the enwiki article about Montenegrin (Q8821).
--- Jura 11:54, 25 January 2018 (UTC)

I've created T185800 for this and patch will follow soon. But before it can be merged, approval by LangCom is needed. Mbch331 (talk) 10:30, 27 January 2018 (UTC)
@Millosh: what do you think ? It's a request for a language code at Wikidata (not a new wiki/interface language).
--- Jura 17:20, 28 January 2018 (UTC)
It's a valid language code (per [1]) and I see no reason why it shouldn't be added. --Millosh (talk) 19:14, 2 February 2018 (UTC)
Is that a personal statement or an official statement from the langcom? If the latter, please add a statement to phab:T185800. Mbch331 (talk) 20:15, 2 February 2018 (UTC)
@Mbch331: thanks for the patch. I think you can go ahead with it @Millosh: is the m:Language_committee#Current member knowledgeable about this language (or closely related ones). --
--- Jura 07:20, 3 February 2018 (UTC)
@Jura1: My part of the work is done anyway. I don't have the right to merge patches. That's only to the WDME devs. I'm just a volunteer dev that submits patches. Mbch331 (talk) 07:23, 3 February 2018 (UTC)
Tracked in Phabricator
Task T124758
We're still working on solving this issue. It's a tough one. Lea Lacroix (WMDE) (talk) 12:24, 24 April 2018 (UTC)

Let's get this section archived. There is nothing won by keeping it here. The ticket is open. --Lydia Pintscher (WMDE) (talk) 12:42, 27 January 2019 (UTC)

  • It somehow illustrates a lack of feedback to the community on arcane interna in the development processes. Can we have a least a summary understandable by a random contributor about what needs to be done and how long that may take? --- Jura 13:14, 27 January 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Patch for T185800 has been merged. Next passing of the train it will be live. Mbch331 (talk) 18:42, 15 March 2018 (UTC)

Placement in identifier section

To detail a point brought up earlier, is it possible to add an option to place statements of a few properties (e.g. P1036, sample use: Q64#P1036) in the identifier section of pages (in the sample Q64#identifiers). Currently the sort seems to be based exclusively on datatype.
--- Jura 06:52, 17 February 2018 (UTC)

Why not simply change P1036 to external-id datatype? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:51, 17 February 2018 (UTC)
Dewey Decimal Classification (P1036) is not an identifier but classifier. An identifier identifies an unique instance of class. In contrast, a classifier is used to tag all instances of a class. For example the code 813.54 means American narrative prose from 1945 to 1999. If P1036 is an identifier, we would add that code to the item about the class of American narrative prose from 1945 to 1999. But since P1036 is used as classifier, we add the code to all instances of that class. --Pasleim (talk) 14:10, 17 February 2018 (UTC)
would be more useful to have such classes on Wikidata and to use this property only on the class item. This would allow us to encode the class definition as statements in Wikidata and to infer instances automatically, for example with {{Implied instances}}. Or to add automatically the statements in the class item to the explicit instances of it without having to put the logic into the bot (or an out wiki inference rule). author  TomT0m / talk page 14:17, 17 February 2018 (UTC)
In one sense, certainly; but we have plenty of other external-id properties on Wikidata that "identify" classes of things. Do we really add Dewey codes to every item about a book with that code? The examples on the property page certainly do not suggest that. I also note that, in the property proposal, User:Merrilee of OCLC said "The Dewey identifiers are just that -- identifiers... I had an in depth conversation with the Dewey Editor, Michael Panzer, last week and this use of Dewey identifiers as free for all to use lines up with his understanding as well.". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:17, 17 February 2018 (UTC)
  • I think the datatype of one or the other properties for which this can apply should be discussed elsewhere. I think there is a general need for this.
    --- Jura 08:24, 20 February 2018 (UTC)

Can you please formulate a clear request that is left over here now? I am a bit lost as to the exact issue we're still trying to solve. --Lydia Pintscher (WMDE) (talk) 12:45, 27 January 2019 (UTC)

  • Whatever their datatype, it should be possible to place P1036 and P2888 in the identifier section (i.e. under the Heading "identifiers" on entities). --- Jura 13:16, 27 January 2019 (UTC)

Formatter url and string datatype

A problem brought up further up on this page. It seems that this works sometimes and sometimes it doesn't. What's is the development status on this? I think ideally this would work as for "external id"-datatype when the url is set on a property.
--- Jura 08:30, 20 February 2018 (UTC)

I checked and it is possible to do. Can you give me 2 or 3 examples where this would hold and not be an external identifier? I need to look into some of the details because of the RdF export before someone can work on it. --Lydia Pintscher (WMDE) (talk) 10:59, 15 March 2018 (UTC)
Full list: https://query.wikidata.org/#SELECT%2a%7B%3Fp%20wikibase%3ApropertyType%20wikibase%3AString%3Bwdt%3AP1630%5B%5D%7D
--- Jura 11:41, 15 March 2018 (UTC)
Ok should be doable. Can you open a ticket for it? Or Léa? --Lydia Pintscher (WMDE) (talk) 14:34, 13 April 2018 (UTC)
Léa? That would be helpful. Apparently, I can't write tickets. Besides, I surely wouldn't find it if someone had already opened one.
--- Jura 14:53, 13 April 2018 (UTC)
done, phabricator:T192188
Tracked in Phabricator
Task T192188
--Pasleim (talk) 00:25, 14 April 2018 (UTC)

Let's get this section archived. There is nothing won by keeping it here. The ticket is open. --Lydia Pintscher (WMDE) (talk) 12:41, 27 January 2019 (UTC)

I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --Lydia Pintscher (WMDE) (talk) 08:26, 23 April 2018 (UTC)
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Contact_the_development_team/Archive/2019/01&oldid=889128613"