This message was deleted.
# cachings
Slackbot
03/07/2023, 2:22 PMThis message was deleted.
v
Vampire
03/07/2023, 2:29 PMJust in case the answer is no, you could have the validation task not fail but write the result to some file and have a finalizedBy task that evaluates the result file and fails. Then it should be cached. :-)
m
melix
03/07/2023, 2:29 PMyes, but it's a bit cumbersome, since it moves the failure out of its context
l
lptr
03/08/2023, 9:36 AMWhy would you re-execute a failed task with the exact same inputs?
m
melix
03/08/2023, 9:36 AMbecause CI
melix
03/08/2023, 9:37 AMthere are different pipelines, they would benefit from sharing the failure
melix
03/08/2023, 9:37 AMthe verification task can be slow, so it's better if we don't execute it if we know it's going to fail
l
lptr
03/08/2023, 9:39 AMOne other way would be to separate the slow part out into its own cacheable task. I assume that’s where you collect all the dependencies and their versions. That task could always succeed, and produce an output. Then you could validate that output quickly.
m
melix
03/08/2023, 9:40 AMyes, that's the solution Vampire suggested, and that I said is a bit annoying because it moves the failure out of its context. But I could do that.
l
lptr
03/08/2023, 9:47 AMMy suggestion is slightly different in keeping the validate task failing, so the failure is not out of context.
m
melix
03/08/2023, 9:48 AMright, slight variation indeed
3 Views