<@U06A03HV1> remind me, is the only reason we don'...
# development
b
@witty-crayon-22786 remind me, is the only reason we don't materialize big files somewhere in the pants cache because it'll grow without bound?
w
yes
b
I'm thinking we'll need a middle ground for it, because per daemon is too terse
w
i could believe that it exercises our clean shutdown code too aggressively to create and remove all of those, yea. but keep in mind that all of the sandboxes are also created/deleted that way
so it’s not clear how the timing would be different in this case. maybe it’s actually that we are not making teardown of those directories async the way we make teardown of processes async, and so the daemon is taking long to restart?
(i mean… you can’t really make the teardown of the daemon async)
b
Well I'm also thinking there's benefit in this outlive daemon runs
w
for sure. i’m just trying to first and foremost reason through how it might be connected to the flakiness
b
Yeah next week I'm hoping to tease that out
w
thanks again for all your work on this.
b
It's still ongoing apparently 🥲😅🤣
💜 3