Wikidata:Contact the development team/Archive/2014/11

This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

wbsetclaimvalue

Someone can check if API wbsetclaimvalue work? I tried with example data, claim=Q4115189$E3363DF2-8230-4F05-B42A-57CB5AA61D04&value={"entity-type":"item","numeric-id":1}&snaktype=value&token=cf874a0b42c245e6a513bbedd1b784e9+\, but don't works, return error: <error code="noclaim" info="The claim parameter must be set" xml:space="preserve">See https://www.wikidata.org/w/api.php for API usage</error> . --ValterVB (talk) 19:26, 23 October 2014 (UTC)

@Lydia Pintscher (WMDE): or someone of the staff: can you check this? Thanks. --ValterVB (talk) 18:51, 28 October 2014 (UTC)

@Hoo man: Can you have a look maybe? --Lydia Pintscher (WMDE) (talk) 20:04, 28 October 2014 (UTC)

Looking at this, please double check you are actually sending the claim parameter along. The actual paramter presence checking happens in a method defined by MediaWiki that's very widely used, so I'd be surprised if it didn't work. - Hoo man (talk) 07:49, 29 October 2014 (UTC)

@Hoo man: Sorry for late replay. Solved, The problem was that I have encoded also claim= and not only the claim value. --ValterVB (talk) 08:31, 2 November 2014 (UTC)

Error when adding incorrect sitelinks

Reproduce:

1. Try to add aa:Test to Q4115189
2. Shows "The external client site did not provide page information"
3. Then click "Save" or "Cancel"
4. Not able to edit this item.

--GZWDer (talk) 12:29, 31 October 2014 (UTC)

This is expected behaviour as aa:Test doesn't exist, thus the external client site (which is aa.wikipedia.org in this case) can't provide information about it. - Hoo man (talk) 00:21, 1 November 2014 (UTC)

@Hoo man: Expected behavior: After pressing "Save" error will occur again; After pressing "Cancel" return to view mode.

Actual behavior: After pressing "Save" "Saving…" displays and no response; After pressing "Cancel" the Cancel button become disabled, and Save still enabled; not able to edit other part of the item.

--GZWDer (talk) 04:20, 1 November 2014 (UTC)

Sorry, I apparently didn't read this carefully enough. I've just checked this and it seems to have been fixed on master (I can reproduce it on Wikidata, but not on Wikibase master). Cheers, Hoo man (talk) 17:32, 1 November 2014 (UTC)

RDF Dumps broken (?)

I asked on IRC about a problem with the RDF Dumps:

2014-10-24 06:11:52 <nloth> hi. I have a question about the Wikidata RDF dumps

2014-10-24 06:13:08 <nloth> I'm getting an error: illegal escape sequence value: " (0x22) when I try to load it (into Jena)

2014-10-24 06:13:17 <nloth> org.apache.jena.riot.RiotException: [line: 142128, col: 121] illegal escape sequence value: " (0x22)

2014-10-24 06:13:39 <nloth> line 142128 looks like this: <http://www.wikidata.org/entity/Q16873>; <http://www.wikidata.org/entity/P18c>; <http://commons.wikimedia.org/wiki/File:\"Retrat_de_l'escriptor_Juan_Carlos_Onetti_(1909-1994)\".png> .

2014-10-24 06:14:00 <nloth> Reading the RDF spec http://www.w3.org/TR/n-triples/#n-triples-grammar

2014-10-24 06:14:17 <nloth> I see [9] STRING_LITERAL_QUOTE ::= '"' ([^#x22#x5C#xA#xD] | ECHAR | UCHAR)* '"'

2014-10-24 06:14:49 <nloth> To me that looks like "R might be an illegal sequence?

2014-10-24 06:17:35 <nloth> hmm.. no one around?

2014-10-24 06:17:57 <GerardM-> nloth the latest ones are all corrupted due to a change in the JSON format

2014-10-24 06:18:11 <GerardM-> they did not fix it. I am not aware that they know how to

2014-10-24 06:18:37 <nloth> The RDF dumps are corrupted coz of a change in the JSON format?

2014-10-24 06:18:39 <GerardM-> both for the future dumps and for the ones that exist

2014-10-24 06:18:57 <nloth> ok well that explains it :(

2014-10-24 06:19:27 <nloth> thanks

2014-10-24 06:21:45 <GerardM-> yeo

2014-10-24 06:21:45 <GerardM-> yep

2014-10-24 06:22:14 <GerardM-> as a result there are no latest statistics for Wikidata

Is this a confirmed issue? Is there a Bugzilla link where I can find out more about what it would take to fix it?

@Daniel Kinzler (WMDE): Can you have a look please? --Lydia Pintscher (WMDE) (talk) 20:06, 28 October 2014 (UTC)

Just discussed this. Do you mean the RDF dumps created by Markus Krötzsch? You'd have to contact him for that. --Lydia Pintscher (WMDE) (talk) 13:49, 6 November 2014 (UTC)

b flag in WP´s Watchlist?

Edits made here at Wikidata is in Wikipedia marked as D in my Watchlist. But I see that edits from flooders are not marked as a b-edit there. The edits in Q18451092 by "User:GZWDer (flood)" are hidden when I request bots edits to be hidden in my watch-list, but I fail to to see any b in that line as long the edit is visible. I only see a D. -- Innocent bystander (talk) 15:46, 4 November 2014 (UTC)

This is bugzilla:60746, which needs more input from people. --Lydia Pintscher (WMDE) (talk) 14:19, 7 November 2014 (UTC)

Problem of claim reordering

Reproduce:

1. Reorder a claim to other place
2. Try to reorder other claim, able to save but in fact not saved
3. Try to reorder the same claim as step 1, succeeded
4. Fresh the page, then able to reorder another claim, but also only one

Browser same as above.--GZWDer (talk) 10:52, 6 November 2014 (UTC)

Yeah ordering is really broken. We need to rework that whole system as part of the UI redesign. Tracking for that is at bugzilla:58180. --Lydia Pintscher (WMDE) (talk) 14:29, 7 November 2014 (UTC)

Can't change rank only 2

Reproduce:

1. Trying to change the rank of the P17 claim in Q4115189, "Save" is blue but after clicking, no change is saved, the rank reverts after freshing the page
2. Change rank and value works [1]
3. Change rank and order works [2]

See Wikidata:Contact the development team/Archive/2014/06#Can't change rank only. My browser: 5.0 (Windows NT 5.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/31.0.1650.63 Safari/537.36 SE 2.X MetaSr 1.0--GZWDer (talk) 10:42, 6 November 2014 (UTC)

Also reproduced in: MSIE 8.0; Windows NT 5.1; Trident/4.0;--GZWDer (talk) 10:44, 6 November 2014 (UTC)

A fix for that is coming soon. --Lydia Pintscher (WMDE) (talk) 14:25, 7 November 2014 (UTC)

For the same reason, both coat of arms entries in Q184 have normal rank, but if you query with the API one is preferred. There is discrepancy between the UI and the API result. The change with which I made it preferred doesn't show up in the history. --JulesWinnfield-hu (talk) 11:40, 6 November 2014 (UTC)

Hmmm I can't see the issue. Are you still seeing it? Was there some caching issue maybe? --Lydia Pintscher (WMDE) (talk) 14:25, 7 November 2014 (UTC)

Yes. @Lydia Pintscher (WMDE): If you can not reproduce it, please post diff with changing rank only here. By UI, not API.--GZWDer (talk) 15:42, 7 November 2014 (UTC)

I can reproduce that part. As I said above that'll be fixed soon. What I can not see is the discrepancy between what is stored and what the UI shows that Jules mentioned. --Lydia Pintscher (WMDE) (talk) 15:57, 7 November 2014 (UTC)

The API result changed back to normal. Now is preferred again, because I changed it again. If you see normal, change the first entry to preferred and view the API result link. --JulesWinnfield-hu (talk) 16:26, 7 November 2014 (UTC)

Ahhh ok. Now I can reproduce it as well. Thank you. Looking into it. Might be some weird caching issue. --Lydia Pintscher (WMDE) (talk) 15:20, 10 November 2014 (UTC)

Ok we looked into it some more. The issue is caused by the same issue that caused the UI not showing the change after reload. A fix for both is going to go live in 1 week. bugzilla:73258 and bugzilla:73257 are additional things we can do to guard against this in the future.

Thanks for finding and reporting this. --Lydia Pintscher (WMDE) (talk) 10:20, 11 November 2014 (UTC)

Sequences

Note the question on Wikidata:Project chat#Sequences. Jeblad (talk) 08:35, 11 November 2014 (UTC)

Error on valid language code

Tried to add "Vuona gånågisrijkka" with language code smj for Norway (Q20), this is the name in w:en:Lule Sami Language according to w:nn:Noreg, but got an error on the language code. The code is in the drop down list for valid languages. This language does not have a valid interface language. Jeblad (talk) 15:29, 26 October 2014 (UTC)

This is already being tracked in bugzilla. --Lydia Pintscher (WMDE) (talk) 20:07, 28 October 2014 (UTC)

Yes, I added bug 72590 - Monolingual code is missing for Kvensk (fkv), Romani (rom) and Scandoromani (rmg-variant). There is also another bug 72126 - monolingual text does not accept sr-cyrl and a number of other language codes. I think it is important to fix those bugs, and it should be fixed by allowing use of the additional languages for mono-/multilanguage codes and not by limiting them to the valid codes for localization. Those code sets are not the same and should probably not be interpreted as the same. The first one is an approximation to all possible languages, which is much larger than the valid codes for localization of the software. Actually the mono-/multilanguage codes does not imply localization at all. Jeblad (talk) 02:29, 12 November 2014 (UTC)

Sitelinks

If anyone has time I would like som input on Wikidata:Requests for comment/Sitelinks with fragments. There is also a thread at Wikidata:Project chat#One vs. several sitelink-item correspondence. Jeblad (talk) 08:47, 11 November 2014 (UTC)

I have said several times that I think this is a very bad idea. It will break important assumptions we're making about the world and it'll take significant development time to work around those for questionable benefit. --Lydia Pintscher (WMDE) (talk) 09:17, 11 November 2014 (UTC)

Please note the Hash URIs as described by W3,[3] and also note that use of a hash URI does not break the assumption of unique URIs. The only thing I am aware of that will break is the current assumption that the hash (fragment) is not part of the URI. If there is other assumptions that will break feel free to give examples and links to docs or code. Jeblad (talk) 09:53, 11 November 2014 (UTC)

There is one assumption that could create a problem, and that is the unique backrelation in an external page. Hash URIs will make it possible to have two items point to the same page, which makes it necessary to return and handle two structures. This can be counteracted by giving a hash-less URI priority, that is simply listing it first. Jeblad (talk) 10:32, 11 November 2014 (UTC)

Can we have a page, possibly a subpage under mw:Wikibase, that is linked from Help:Sitelinks, that list all the assumptions? The same thing should probably also be done for other critical assumptions made by the developers. Jeblad (talk) 03:30, 12 November 2014 (UTC)

Adding links through wikipedia

When one links a new page on a wikipedia to another new page one a wikipedia, both not yet linked to wikidata, then a new item is created and for both languages a label is created. However, if one of the pages is linked to wikidata, the added sitelink is not accompanied with a label. Can that be solved? Lymantria (talk) 13:43, 7 November 2014 (UTC)

@Hoo man: What do you think? --Lydia Pintscher (WMDE) (talk) 14:31, 7 November 2014 (UTC)

We had this discussion before, but decided that the Wikibase software shouldn't make the assumption that labels equal the page title. For item creations we decided to set labels though, as items without any labels are extremely hard to identify (thus the bad assumption is bearable there compared to having these unidentifiable items). Cheers, Hoo man (talk) 15:37, 7 November 2014 (UTC)

For wikipedia-users it is hard to understand in which case they have to add a label and in which case not. Perhaps you could reconsider to add the label equal to the page title. Lymantria (talk) 10:39, 10 November 2014 (UTC)

Good point. Will discuss this with Marius next week some more. --Lydia Pintscher (WMDE) (talk) 13:03, 13 November 2014 (UTC)

cross wiki authentication does not work

see Wikidata:Forum#Verlinkung_mit_WD_geht_nicht.2C_authentication_schl.C3.A4gt_fehl (German, but if you need English, I will gladly translate) --Herzi Pinki (talk) 00:28, 13 November 2014 (UTC)

Thanks. Will answer there. --Lydia Pintscher (WMDE) (talk) 12:55, 13 November 2014 (UTC)

Cannot edit Q14920473

No edit buttons anywhere on ship grounding (Q14920473) for me, neither in Firefox nor Chrome. Only for that item?! Emptying the cache had no effect. Anyone else with this problem? -- LaddΩ chat ;) 18:34, 14 November 2014 (UTC)

Came back with https://www.wikidata.org/w/index.php?title=Q14920473&action=purge --JulesWinnfield-hu (talk) 19:02, 14 November 2014 (UTC)

The UI is gone. Can't add statements. --JulesWinnfield-hu (talk) 19:27, 14 November 2014 (UTC)

I get Javascript error also. undefined is not a function when click on add to add statements. --JulesWinnfield-hu (talk) 19:33, 14 November 2014 (UTC)

thanks Jules, that item cache purge seems to have resolved it for me. -- LaddΩ chat ;) 21:21, 14 November 2014 (UTC)

Not receiving reversion notifications

I am not receiving reversion notifications through web, although they are enabled in my preferences. Here is one of the reversions I have not been notified of. Petr Matas 07:51, 18 November 2014 (UTC)

Did you receive them in the past on Wikidata? --Lydia Pintscher (WMDE) (talk) 09:27, 18 November 2014 (UTC)

I have never received any and I have never changed the notification preferences. Petr Matas 10:05, 18 November 2014 (UTC)

That's what I feared. Thank you. I will investigate. Echo probably doesn't properly recognize the revert. --Lydia Pintscher (WMDE) (talk) 12:00, 18 November 2014 (UTC)

I have filed bugzilla:73562 for it. --Lydia Pintscher (WMDE) (talk) 16:37, 18 November 2014 (UTC)

Thanks, Lydia. Does it also affect other users? Is there any way around this? Petr Matas 16:50, 18 November 2014 (UTC)

I just tested it for myself and I can reproduce it. I don't know if others are affected but I assume it is the same for everyone. --Lydia Pintscher (WMDE) (talk) 16:55, 18 November 2014 (UTC)

Kunal just made a patch for this. Still needs to go through review but should become available with the next deployment in 2 weeks if nothing goes wrong. --Lydia Pintscher (WMDE) (talk) 21:53, 18 November 2014 (UTC)

Some statements incorectly show their value is a deleted item

I have just created some statements: [4], [5], [6]. Their values are shown as pointing to deleted items, but those items do exist. The first two cases point to an item I have just created. The third case points to an item, which existed since long time ago. Petr Matas 20:08, 18 November 2014 (UTC)

Marius is looking into it right now. It seems we screwed that up in the last deployment. If there is no label it is shown as deleted which is obviously wrong. Thanks for poking. --Lydia Pintscher (WMDE) (talk) 20:43, 18 November 2014 (UTC)

Tracking at bugzilla:73580. --Lydia Pintscher (WMDE) (talk) 20:53, 18 November 2014 (UTC)

Aaaand fixed again. --Lydia Pintscher (WMDE) (talk) 21:52, 18 November 2014 (UTC)

Wow, that was incredibly fast! Petr Matas 22:40, 18 November 2014 (UTC)

Symmetric properties

Are there any plans to implement anything like mentioned in Wikidata:Requests for comment/Make Symmetric & Inverse Property addition automatic and bidirectional ? --DixonD (talk) 20:10, 19 November 2014 (UTC)

There will be no automatic addition of symmetric or inverse properties, no. However something the student team that is working on constraint violation reports is thinking about is prompting the user to add the other statement as well. --Lydia Pintscher (WMDE) (talk) 08:47, 20 November 2014 (UTC)

Automatically restore links?

When a page is deleted on a client (Wikipedia, etc) its sitelink is now removed automatically from its Wikidata item. However, some page deletion is because of merging page history, or selective undeletion (Although there're MergeHistory and RevDelete), or to spilt page history. I think when a page is deleted on a client, its sitelink should be restored automatically from its Wikidata item, unless the item is deleted. Otherwise, admins must find the item the page was linked to. (Probably related to https://old-bugzilla.wikimedia.org/show_bug.cgi?id=54162 )--GZWDer (talk) 12:01, 23 November 2014 (UTC)

Good point. I am not sure if we can do this but I opened phabricator:T75908. --Lydia Pintscher (WMDE) (talk) 18:26, 25 November 2014 (UTC)

Rendering of coordinates

Tracked in Bugzilla
Bug 73601

Some coordinates are rendered as 0°0'NFF"N, 0°0'NFF"E. Is that bug or feature? --Jklamo (talk) 16:03, 23 November 2014 (UTC)

Looks like all claims with a missing precision value are affected. --Pasleim (talk) 13:12, 24 November 2014 (UTC)

This is bugzilla:73601, thank you. --JanZerebecki (talk) 18:23, 25 November 2014 (UTC)

Wikidata:Do not merge

I can't seem to open this page. It gives me Wikimedia Error whenever I try to open it. 159.92.9.130 15:13, 25 November 2014 (UTC)

A split into subpages is in preparation. See Wikidata talk:Do not merge. --Succu (talk) 15:27, 25 November 2014 (UTC)

Bug in WDQ?

Why is it that Query: claim[361:18552670] does not return Bunker Battery Den Hoorn (Q17460014)? Petr Matas 22:07, 24 November 2014 (UTC) (copied from Wikidata talk:List of queries#Bug?)

@Magnus Manske: Any idea what's wrong? --Lydia Pintscher (WMDE) (talk) 18:32, 25 November 2014 (UTC)

Honestly, no idea. The item shows up for other queries, just not this one. The next edit, or maybe even just a purge, will fix this, but I'm not sure why it doesn't show up. Will keep an eye out. --Magnus Manske (talk) 19:02, 25 November 2014 (UTC)

OK, I made an edit, and it picked it up. Still not sure why the part of (P361) one wasn't picked up. --Magnus Manske (talk) 19:16, 25 November 2014 (UTC)

Problems to insert qualifiers

I can no longer insert third and fourth qualifiers in Peter-Klaus Schuster (Q2073454). Is there a bug?--Oursana (talk) 11:42, 25 November 2014 (UTC)

Gerrit change 175701 will fix this. Cheers, Hoo man (talk) 14:27, 25 November 2014 (UTC)

This changed has been deployed now, so you should be able to add qualifiers again. Thank you for letting us know. Hoo man (talk) 20:37, 25 November 2014 (UTC)

Thanks for your quick help--Oursana (talk) 22:17, 25 November 2014 (UTC)

Dimensions and units for the quantity datatype

Is Wikidata:Requests for comment/Dimensions and units for the quantity datatype to be interpreted as a change request for the quantity datatype, or is it a list of wanted properties, or is it a list of example units to be used for properties? I am not sure how to interpret the list. Dimensions can say something about which units can be transformed into which other ones, but it is a feature that is very difficult to model.

For example you can say that something is a measure of length, like arrow shot and stone throw (old Norse length measures), but you can not transform some specific length unit into the SI-value meter. Still you can transform from one length unit at that era into another length unit from the same era, there are 4 stone throw on one arrow shot, or you can transform with low accuracy and use the approximation that 1 stone throw is about 35-75 meter. There are also cases where you don't know which version of a length unit is used, even if it reads foot or alen.

I am not sure how to model dimensions with known errors or uncertainty, even if I think that it is possible. The general idea is to say that you have transform rules between some quantities but not between all of them, and that missing transform rules will create isolated islands where you only can transform the quantities within those that define the island. More generally you can say that valid transformations exists if the error bound they produce is within your accepted range.

The previous makes me think that the linked RFC is only talking about a list of transformable example units to be used for properties, it is not talking about a new implementation for quantities within a dimension. Is this a correct interpretation? Jeblad (talk) 03:13, 12 November 2014 (UTC)

The page was for collecting units that should be handled first. The list got way too large though. --Lydia Pintscher (WMDE) (talk) 13:02, 13 November 2014 (UTC)

Thanks, then it was as I suspected. Jeblad (talk) 20:26, 13 November 2014 (UTC)

Jeblad: The articles you link to give estimates in meters of the value of an arrow shot and of a stones throw and how these varied in different places at different times so we will have arrow shot (Sweden 1600) = 200m +- 20m. which we should be able to perfectly well express semantically. I don't think there is a problem that we cannot solve here. Filceolaire (talk) 20:21, 16 November 2014 (UTC)

Sorry, but which arrow shot do you refer to? Which era, and which geographic region? Or do you refer to the attempts to model the actual length? There are models that estimates the arrow shot to within +/- 0.01 meter, but I don't think anyone would say the estimate is anywhere near accurate. An arrow shot is usually just said to be somewhere between 140-300 meter, but we know that there are 4 stone throw on one arrow shot. That is, within the given measurement system there are exactly 4 stone throws on one arrow shot, even if we can't accurately relate this to the SI-system. Jeblad (talk) 12:13, 17 November 2014 (UTC)

Er, era: 1600, region: Sweden. That is what the bit in brackets means. We will have a dozen different arrow shot units each corresponding to a different era and region just as we will need to have a dozen different ounce and pound and acre units corresponding to different eras and regions if we ever start to codify the historic measurement units used in other places. Filceolaire (talk) 01:42, 18 November 2014 (UTC)

I personally would be glad if we had some datatypes for SI-meassures like meter, kilogram etc. Probably we will have some datatype where you can add every kind of units regardless how they fit into the SI-System like sqarefoot by arrow shot or horsepower by pound. So you might enter every possible unit, but don´t get it transformed into SI-meassures for now. Later it my be replaced by some specific datatype that can handle local/historical units, but this is no priority. It just makes no sense to discuss historic acres and arrow shots at this stage of development.--Giftzwerg 88 (talk) 11:28, 30 November 2014 (UTC)