chore: improve error of download config #4603
Conversation
maan2003
force-pushed
the
ma/jj-qquzrwnnkutr
branch
from
ea440fb
to
0dc05ee
Compare
maan2003
force-pushed
the
ma/jj-qquzrwnnkutr
branch
from
0dc05ee
to
433d53e
Compare
- use it in download client config
maan2003
force-pushed
the
ma/jj-qquzrwnnkutr
branch
from
433d53e
to
da6cca6
Compare
now returns the correct error with added context
maan2003
force-pushed
the
ma/jj-qquzrwnnkutr
branch
from
da6cca6
to
7c135da
Compare
| op_name: impl Into<String>, | ||
| op_fn: F, | ||
| interval: Duration, | ||
| max_attempts: u32, |
There was a problem hiding this comment.
max_attempts < timeout , just saying.
Isn't this poll from devimint where I just added deadlines and turned interval to be growing? Can we move it here from there and rename?
There was a problem hiding this comment.
That one uses ControlFlow, while here we want that to be implicitly implied by Result. retry could be implemented using poll though.
There was a problem hiding this comment.
That control flow there is a bit of a PITA, BTW (because of type inference on two variants). Not sure how badly we need it. In testing if there's an error that should really terminate polling... maybe we should just panic, instead of complicating common case? Just thinking aloud.
But one way or the other - would prefer to see a timeout/deadline approach here.
There was a problem hiding this comment.
Will leave it to @maan2003, don't have a strong opinion here.
There was a problem hiding this comment.
🤷 devimint and this retry can just stay separate
attempt + interval is pretty close to deadline though
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
|
Let's land it already, doesn't matter all that much in a great scheme of things, Manmeet is free to follow up (or not). :D |
now returns the correct error with added context