Sophox / sophox

A collection of services exposing OSM data, metadata, and other microservices
Apache License 2.0
31 stars 5 forks source link

Many subjects are missing #27

Closed 1ec5 closed 6 months ago

1ec5 commented 3 years ago

Despite being more or less caught up to the latest OSM and OSM Wiki changes, Sophox is missing a lot of OSM elements and wiki data items. For example, this query says there are only 439 man_made=flagpole flag:wikidata nodes, none of them older than March 8 of this year, but there should be over 6,200 nodes, most of them older than March. It isn’t very consistent; @camelCaseNick also found an older way that is missing. This issue also affects the osmd: namespace: it seems to only be pulling in data items that were edited fairly recently.

/cc @nyurik

1ec5 commented 3 years ago

Is there anything significant about March 11, 2013, and March 7–8, 2021?

So far I haven’t found a tagged but unedited way with an ID below 209,385,395. For example, 209,385,391 is missing. So is 209,385,392 despite being edited in June 2019. 209,385,393 is present, but it was edited on March 17, 2020. 209,385,394 is also present, but it was edited in 2019.

Similarly, node 8,495,271,663 is present, but 8,493,345,199 is missing. Node 1,231,643,461 is also present because it was edited recently.

westnordost commented 3 years ago

Can confirm. This query f.e.

SELECT ?value WHERE {
  ?osm osmt:operator "HUMANA Kleidersammlung GmbH".
}

(https://sophox.org/#SELECT%20%3Fvalue%20WHERE%20%7B%0A%20%20%3Fosm%20osmt%3Aoperator%20%22HUMANA%20Kleidersammlung%20GmbH%22.%0A%7D) results in 9 rows, even though according to taginfo, there should be 19.

1ec5 commented 6 months ago

This issue appears to be fixed:

For example, this query says there are only 439 man_made=flagpole flag:wikidata nodes, none of them older than March 8 of this year, but there should be over 6,200 nodes, most of them older than March.

As far as I can tell, this issue is now fixed. The same query now returns 25,780 results, which is probably consistent with the 24,970 combinations of these tags that taginfo reports, given that Sophox splits statements on semicolons.

Can confirm. This query f.e.

SELECT ?value WHERE {
  ?osm osmt:operator "HUMANA Kleidersammlung GmbH".
}

(https://sophox.org/#SELECT%20%3Fvalue%20WHERE%20%7B%0A%20%20%3Fosm%20osmt%3Aoperator%20%22HUMANA%20Kleidersammlung%20GmbH%22.%0A%7D) results in 9 rows, even though according to taginfo, there should be 19.

There was a slight error in this query. Here’s a fix that now returns 63 results, consistent with taginfo.