Wikidata:Property proposal/Roglo person ID

Roglo person ID edit

Originally proposed at Wikidata:Property proposal/Authority control

Done: Roglo person ID (P7929) (Talk and documentation)

Description	identifier for a person on the Roglo genealogical database
Represents	Roglo (Q83365652)
Data type	External identifier
Domain	human (Q5)
Example 1	Louis XVI of France (Q7732) → p=louis;n=de+bourbon;oc=16
Example 2	Ronald Reagan (Q9960) → p=ronald;n=reagan
Example 3	Jim Morrison (Q44301) → p=james;n=morrison
Number of IDs in source	7.948.574
Expected completeness	always incomplete (Q21873886)
See also	FamilySearch person ID (P2889), genealogics.org person ID (P1819), WikiTree person ID (P2949), etc.

Motivation edit

Genealogical database with almost 8 million unique entries which allows advanced relationship computing (2,388 relationships between Louis XVI of France (Q7732) and Jim Morrison (Q44301) for example). Ayack (talk) 17:01, 21 January 2020 (UTC)[reply]

Notified participants of WikiProject Genealogy

Discussion edit

@Peter17: provided a script to extract data from this website.--GZWDer (talk) 08:52, 22 January 2020 (UTC)[reply]

Support: sounds useful. Nomen ad hoc (talk) 18:55, 22 January 2020 (UTC).[reply]
Support. Please note that the script is just a proof of concept. Peter17 (talk) 21:57, 22 January 2020 (UTC)[reply]
Support but please add a regex to exclude numeric identifiers as they are not stable _at all_. Too bad, a stable identifier is not maintained when following link from one person to another. There is a "Permanent link" button but it is at the bottom of the (possibly long) page. --Melderick (talk) 20:06, 27 January 2020 (UTC)[reply]
- I was wondering why these weren't being used. What happens once they get several ["James" and "Morrison"] (3rd sample above). Do they get changing ";oc="-integers? --- Jura 09:50, 29 January 2020 (UTC)[reply]
  - Yes: http://roglo.eu/roglo?lang=fr;p=james;n=morrison;oc=2;. Ayack (talk) 19:26, 29 January 2020 (UTC)[reply]
    - With "changing", I mean it isn't stable either. --- Jura 08:07, 30 January 2020 (UTC)[reply]
Oppose problem with stability. --- Jura 10:29, 3 February 2020 (UTC)[reply]

@Jura1: Sorry, but I don't understand why you consider that these IDs are not stable. When several persons have the same name, the first to be created as an ID like "p=first name;n=last name", the second "p=first name;n=last name;oc=2;", the third "p=first name;n=last name;oc=3;", etc. These ID don't change later and are considered by Roglo as permanent contrary to numeric IDs. When you click on the "Permanent link" button, you get this ID. Ayack (talk) 11:01, 3 February 2020 (UTC)[reply]

@Jura1: Could you explain you position please? Otherwise, I'll mark again this proposal as "Ready". Thanks. Ayack (talk) 14:23, 7 February 2020 (UTC)[reply]

@Jura1: I agree with Ayack. As far as I can see, the form "p=;n=;oc=;" is stable. For example, since January 30th the numeric identifier has changed at least once, yet all the examples, including the one for Louis XVI with oc=16, are still working properly. So I wonder where you see a stability issue. --Melderick (talk) 07:49, 8 February 2020 (UTC)[reply]

I think the question is what happens when an additional person with the same name is added. Does it renumber everybody? --- Jura 09:54, 8 February 2020 (UTC)[reply]

No, the new one has the same ID as the precedant one with an occurence number +1.

For example, if there are three John Doe, "p=john;n=doe;", "p=john;n=doe;oc=2;" and "p=john;n=doe;oc=3;", the fourth one will be "p=john;n=doe;oc=4;" and the existing ones won't be changed. Ayack (talk) 11:38, 10 February 2020 (UTC)[reply]

that we have unique identfiers is important. I suggest that someone who can speak french fluent explain in the discussion forum what we plan to do and the importance that we have an unique/persistent id - Salgo60 (talk) 23:49, 18 February 2020 (UTC)[reply]

I propose the following regex to validate the identifier : p=[^;]*;n=[^;]*(;oc=\d+)? --Melderick (talk) 08:06, 8 February 2020 (UTC)[reply]

Notified participants of WikiProject Genealogy @Ayack, GZWDer, Peter17, Jura1: Done Roglo person ID (P7929) —Eihel (talk) 23:06, 18 February 2020 (UTC)[reply]

~~ATTENTION~~

~~As the sign "=" does not work in formatter URL (P1630), please note the following:~~

The root of the URL does not change: http://roglo.eu/roglo?

Previously, the identifier was p=FIRST_NAME;n=LAST_NAME and sometimes with a person number p=FIRST_NAME;n=LAST_NAME;oc=NUMBER

The new identifier will be noted as: FIRST_NAME;LAST_NAME;NUMBER

If no number exists, please keep the final ; without anything after: FIRST_NAME;LAST_NAME;

Keep the first and last names of the site for each part of the identifier, example: Louis XVI of France (Q7732) → louis;de+bourbon;16

Notified participants of WikiProject Genealogy @Ayack, GZWDer, Peter17: The new format of IDs is that of the site, described in the examples of the proposal. In production. Cordially. —Eihel (talk) 20:43, 19 February 2020 (UTC)[reply]

Please re-open the discussion if you think you can't create it as proposed. It's not acceptable to create a property different from what has been discussed. --- Jura 02:19, 19 February 2020 (UTC)[reply]
@Jura1:
- The proposal is applied more than 7 days ago.
- 3 users approve this proposal.
- As you already wrote to me: it is more than enough to create a property. When it affects you personally, we have to stop everything, but when you are asked for patience, you keep going. But there is nothing to complain about.
- Only there, I created strictly what was proposed and the changes I write above are a usual measure. When a URL does not point to the correct page, a third-party URL Formatter is used. For example, this technique has already been used on The Cardinals of the Holy Roman Church ID (P7619) in the same way… and it works perfectly. I would like Epìdosis to confirm my actions as compliant.
- When it comes to questions about the stability, it seems to me that users have already replied you.
Also the discussion is still open. Cordially. —Eihel (talk) 03:28, 19 February 2020 (UTC)[reply]
Its something entirely different you created. If you have a new argument to present, please do so in the discussion, not after having created the property or just before you close the discussion. If you don't want to address the concerns raised, please don't create properties. It's not really helpful if you comment on the messenger instead. As you confirmed that the discussion is still ongoing, I marked the property accordingly: it shouldn't be used in the meantime. I think we should try to find a way to make this link directly. The entire point of linked open data is that it links. --- Jura 03:35, 19 February 2020 (UTC)[reply]
Yes, in fact, as you seem to be saddened by the non-compliance with the proposal, the discussion remains open. But, I repeat, the property is identical to the proposal, except what cannot work otherwise. But once the incident is closed, the discussion will have to be done on Property_talk:P7929. I created this property because the discussion is in favor of this creation, but no one noticed this anomaly. The ArthurPSmith wikidata-externalid-url is made to create a direct link to the external site. Following the property link that I provided in my last message, the proposal for P7619 has been modified accordingly and no one has been offended. You are the only one to "go up to the barricades" and challenge the obvious: identifiers don't work with certain characters. I am in tune with the 3 users who want this property. Cordially. —Eihel (talk) 06:28, 19 February 2020 (UTC)[reply]
So if the agreed solution doesn't work, you just create a property nobody proposed? --- Jura 06:38, 19 February 2020 (UTC)[reply]
@ArthurPSmith: for attention.--GZWDer (talk) 04:22, 19 February 2020 (UTC)[reply]
I think there three options to move forward:

we proceed with the initial proposal and let the gadget do the link. This until Wikibase gets fixed (there seems to be some bug)
we use url-datatype
we try to create three qualifiers, one for each part and generate links with the gadget from that. --- Jura 08:36, 19 February 2020 (UTC)[reply]

@GZWDer: I'm happy to help but I have no idea what's being requested here. Why doesn't '=' work in formatter URL's? It should, I'm sure we have other cases like this. ArthurPSmith (talk) 14:02, 19 February 2020 (UTC)[reply]

@ArthurPSmith: See here, you will find the link does not work.--GZWDer (talk) 14:56, 19 February 2020 (UTC)[reply]

Hello @ArthurPSmith:,
Precisely, I used your tool to counter this problem. It is the succession of parameters (?) containing "=". In my memory, the destination looked like, example: http://roglo.eu/roglo?p%3Dronald;n%3Dreagan and the page result indicated "Pas trouvé: …" (or at least not Ronald Reagan's page). Besides, GZWDer has reverted to an older version of the property (hoping that he has purged the cache), so he can confirm my writings. Finally, the case has already happened and it did not make as much noise. Cordially. —Eihel (talk) 15:08, 19 February 2020 (UTC) Thanks GZWDer ! Yes, "Requête incorrecte" was the result. —Eihel (talk) 15:42, 19 February 2020 (UTC)[reply]

@GZWDer, Eihel: Ah, I see, the UI here is replacing the '=' characters with %-encoding and the target website doesn't handle that correctly. Ok, I modified the testwikidata example so it works. However, somebody seems to have gotten the modified id's working at the current property. Either way I don't think there's anything I particularly need to do here. ArthurPSmith (talk) 18:11, 19 February 2020 (UTC)[reply]

Thank you very much ArthurPSmith. Would that be okay with you, Jura1 and sorry for the inconvenience? —Eihel (talk) 20:39, 19 February 2020 (UTC)[reply]

Looks good. Thanks to everyone involved to make this work. I think we should still file a bug so it might eventually get fixed to work without the redirect service. Wasn't the redirect service meant to be temporary for properties we couldn't transition directly to external-id? If we keep creating more that rely on it, we might not be going in the right direction. @ArthurPSmith: did you have any news recently? --- Jura 10:11, 20 February 2020 (UTC)[reply]

I'm not aware of any progress. T150939 was the main phabricator ticket on this I think, but I don't think it's on anybody's radar right now.

Tracked in Phabricator
Task T150939

ArthurPSmith (talk) 15:08, 20 February 2020 (UTC)[reply]

Oh .. who many redirects does it resolve these days? --- Jura 15:22, 20 February 2020 (UTC)[reply]