[01:15:19] Dunno about stacking jobs, but it might not be that hard to make the job runner itself do the polling and run jobs based on the results [05:05:51] I mean CentralAuth renames effectively do that [05:06:09] And it works terribly and is known to occasionally get stuck and just forget to progress [12:40:19] Introducing: message queues [13:55:25] couldn’t they be queued as some other kafka topic theoretically [13:56:00] I don’t use kafka because it’s really stupid but I’m inclined to make a rabbitmq implementation of jobqueue and eventbus lol [16:53:38] Well Kafka is not a message queue, but I suppose a fork can be used as a spoon in dire circumstances [19:29:21] the wmf sure as fuck want to try and make it one [19:30:44] yeah; its that "everyone is using this, how can we use it" thing developers like to do rather than use something that is actually designed for what they're doing [19:31:44] I probably will switch WO to some rabbitmq based queue instead of the redis one because it's also not really a queue lmao [19:57:47] RabbitMQ is nice and that was my original intention when I was switching from Redis; but it turns out its a bit of a pain. [19:58:12] Stuff like delayed jobs are a pain to implement since they abandoned the OSS plugin and moved it to enterprise [19:58:43] I settled on Apache Artemis eventually which has built in support for "don't release this message to any queue until its time"