When workers claim tasks, TaskTiger moves tasks to a new queue instead of popping them like WakaQ. That means if your task fails with WakaQ then it's just gone.
> how it compares in benchmarks
WakaQ processes many tasks per fork, while I think TaskTiger only processes one? That should make WakaQ slightly faster, but it still depends on too many factors to know without benchmarks.
I actually benchmarked TaskTiger in production a long time ago against Celery. From memory, I think it was slower than Celery. I can't find those benchmarks and even if I could, WakaTime's scale is bigger now so not sure how useful they would be. If you create some benchmarks would love to see them.
> how it compares in benchmarks
WakaQ processes many tasks per fork, while I think TaskTiger only processes one? That should make WakaQ slightly faster, but it still depends on too many factors to know without benchmarks.
I actually benchmarked TaskTiger in production a long time ago against Celery. From memory, I think it was slower than Celery. I can't find those benchmarks and even if I could, WakaTime's scale is bigger now so not sure how useful they would be. If you create some benchmarks would love to see them.