Yurik added a comment.
@Dereckson, this question is only about the default behavior - we obviously
should allow for some "force refresh" method for bots, but it could be some
extra flag like "force=1". My biggest concern is that developers won't use
"allowcaching=1" because they might not
Dereckson added a comment.
I've discussed the issue with French Wikidata users, they expect to be able
to use WDQ as a part of maintenance / mass edit tasks, so a long caching period
would prevent this use.
SLOWVOTE DESCRIPTION
With raising usage for Wikidata we need to have a solution
MichaelSchoenitzer added a comment.
The Gui should have a dopdown to pick a cachingtime (0, 30s, 1m, 5m, 1h, 1
day) and it should default to the same value as the SPARQL endpoint for
consistency reasons.
I would set the default for both to something like 1-10 minutes…
SLOWVOTE
Snipre added a comment.
If there is a way to use uncatched data there is no reason to not optimize
ressources. Documentation about caching duration should be clearly displayed.
SLOWVOTE DESCRIPTION
With raising usage for Wikidata we need to have a solution for caching SPARQL
query
nichtich added a subscriber: nichtich.
nichtich added a comment.
I would also have voted for a medium duration between 5 minutes and 1 hour. The
benefit of caching (faster results) should also be made more clear.
SLOWVOTE DESCRIPTION
With raising usage for Wikidata we need to have a solution
Yurik added a comment.
@Bene, the developer of the tool will have an ability to control caching - by
either forcing freshed data, or being ok with the stale data. The real
question is what is more likely - developer reading documentation to make sure
they add "stale is ok" parameter, or
Bene added a subscriber: Bene.
Bene added a comment.
The gui we provide on query.wikidata.org should always return uncached data
because it is used for testing and showcasing queries.
However, tool developers should care about caching and decide what cache
duration is acceptable for their
Yurik added a subscriber: Yurik.
Yurik added a comment.
Users who need uncached data will always do an extra step (like adding an extra
parameter to the query) to make sure it is not cached. But if we make caching
non-default, fewer developers will bother to check if is cached or not, either