Closed CarbonAlabel closed 5 years ago
Fixed character lp, set to 3600 sec as in CREST.
Fixed corporation wallet journal to be 3600 sec, same as character ones.
What can even change for an alliance's data? Just the executor_corporation_id, and that's not likely done every 5 minutes...
Somewhat similarly, for the character, isn't it just corporation_id and description? CREST's 30mins seems fair.
Decisions time:
I agree with daniel that characters, corporations and alliances just don't change enough to justify a 5 minute timer and the associated cache miss rate, I'm going to leave them all at 1 hour.
Point taken on the implants, I'm going to lower that to two minutes.
The location endpoint is on a 5 second timer so that it can detect people jumping through a wormhole, seeing hatred on the other side and skiddaddling. It's an expensive endpoint for us as a result because people poll it ceaselessly. What's the usecase for online status being 5 seconds?
Attributes change with implants and jump clones, so it might be marginally useful, I don't see why this can't be two minutes as well.
30 seconds is totally adequate for server status.
The corporation journal/transaction information is 1 hour, same as personal. It's also a relatively heavy call. Again, specific use case for it being lower please.
Apparently someone already lowered the implants cache to 5 minutes, but I don't see why it shouldn't be 2 same as clones.
It would be nice if the wallet journal/transaction endpoints could have the cache reduced to allow for real-time isk generation tracking, another example would be the ability to notify a twitch stream that a donation has been made.
Implants and clones changes are deployed.
Twitch stream rewards is a pretty persuasive argument, but I don't know much about CCP policies in that area, I'm talking to people about it.
Shortening the cache time on the wallet endpoints can be discussed in https://github.com/esi/esi-issues/issues/829. As all the other points have either been resolved or given an explanation for, I'm closing this issue.
I've noticed several ESI endpoints whose cache times are inconsistent with similar endpoints and/or direct downgrades from the XML/CREST endpoints they are supposed to replace. Here is a list of them, along with my specific complaints.