Page MenuHomePhabricator

item labels not updating cleanly on Wikidata Sparql service
Closed, DuplicatePublic

Description

It seems that when a label is changed, the value isn't always updated cleanly in the Blazegraph database, which will sometimes return both the old label and the new label.

For example,

PREFIX wd: <http://www.wikidata.org/entity/>

    SELECT ?p ?lang ?o WHERE { 
         wd:Q1006876 ?p ?o .
         BIND (lang(?o) AS ?lang)            
    }

Here for Q1006876 there are three 'en'-language labels returned, "UK borough", "borough in the UK", and "borough in the United Kingdom", when there should only be the latter. This has now been the case for a number of weeks.

Similarly, Q21497886 (label edited this evening) is now also showing two en-language labels

Event Timeline

Jheald created this task.Nov 16 2015, 11:50 PM
Jheald raised the priority of this task from to Needs Triage.
Jheald updated the task description. (Show Details)
Jheald added a subscriber: Jheald.
Restricted Application added a project: Discovery. · View Herald TranscriptNov 16 2015, 11:50 PM
Restricted Application added subscribers: StudiesWorld, Aklapper. · View Herald Transcript
Jheald updated the task description. (Show Details)Nov 16 2015, 11:52 PM
Jheald set Security to None.
Jheald updated the task description. (Show Details)

Probably the same as T116622

It seems there is an issue with updating things in general!

Also now sometimes seeing this for Commons category (P373)

eg this query, for Q7981506:

PREFIX wd: <http://www.wikidata.org/entity/>
PREFIX wdt: <http://www.wikidata.org/prop/direct/>

    SELECT ?item ?commonscat WHERE { 
         BIND (wd:Q7981506 AS ?item) .           
         ?item wdt:P373 ?commonscat .
    }

Should only return "Equestrian statue of the Duke of Wellington, Glasgow"

However, it is also returning "Wellington Monument, Glasgow", a previous value that was changed on 16th November (diff).

It can go away, if the property value is deleted & then re-created -- eg Q17535092, Q17532740, Q17551225, which all did have spurious duplicates, but are now fine since I have edited them a few minutes ago.

Not sure if this is the same issue as the thing that specifically affects labels, or whether this is a different case of updates being dropped (eg when the server was under pressure).