could check if the cover art is already in CAA maybe?
Freso
/query Leftmost
...
chirlu`
SELECT extract_path_value(edit.data, 'link_type/id') AS rt, link_type.link_phrase, edit.status, COUNT(*) FROM edit, link_type WHERE edit.type = 90 AND extract_path_value(edit.data, 'link_type/id')::int = link_type.id AND edit.id BETWEEN 33000000 AND 34000000 GROUP BY rt, link_type.link_phrase, edit.status;
bitmap ^ (untested, but compiles)
regagain joined the channel
bitmap, new attempt with a CTE and a smaller sample: WITH foo AS (SELECT extract_path_value(edit.data, 'link_type/id')::int AS rt, edit.status FROM edit WHERE edit.type = 90 AND edit.id BETWEEN 33900000 AND 34000000) SELECT foo.rt, link_type.link_phrase, foo.status, COUNT(*) FROM foo, link_type WHERE foo.rt = link_type.id GROUP BY foo.rt, link_type.link_phrase, foo.status ORDER BY COUNT(*) DESC;
So, in summary: 31756 edits were applied, 17 were voted down, 108 got cancelled.
2 failed for technical reasons.
Thats 0.05% of the add-relationship edits voted down, 0.34% cancelled.
0.006% failed.
bitmap, without autoedits and a larger sample: WITH foo AS (SELECT extract_path_value(edit.data, 'link_type/id')::int AS rt, edit.status FROM edit WHERE edit.type = 90 AND edit.autoedit = 1 AND edit.id BETWEEN 33000000 AND 34000000) SELECT foo.rt, link_type.link_phrase, foo.status, COUNT(*) FROM foo, link_type WHERE foo.rt = link_type.id GROUP BY foo.rt, link_type.link_phrase, foo.status ORDER BY link_type.link_phrase, COUNT(*) DE
Why not store DB-specific info in a table in individual projects, with editor ID (from MeB) as the key?
Then when the user logs in via MeB, we somehow get the user ID, and the specific site gets the specific info?
chirlu`
LordSputnik: But how are the specific sites notified about removals? The user will never log in again.
Gentlecat
does that mean that we actually don't have to use oauth2 internally?
chirlu`: can check the same table
chirlu`
Gentlecat: What table?
Gentlecat
user table in MeB database
maybe LordSputnik meant something else
LordSputnik
OAuth could be avoided internally if you can access the MeB user table directly in all internal projects
But I think we might as well stick with OAuth
The benefit of storing site-specific data with the sites is that it's potentially less work to implement that than a way of storing and accessing that data on MeB
alastairp
LordSputnik: we basically do already
LordSputnik has quit
Freso
Other people are more than welcome to assist in adding to the notes too. Just let me know your Google id thingies.