Stream: t-compiler/help

Topic: making dataflow public

tmandry (Sep 16 2019 at 21:36, on Zulip):

can someone familiar with dataflow comment on whether it's ok to be made public for clippy?

tmandry (Sep 16 2019 at 21:36, on Zulip):

I've done some work on it and think it's fine, but wanted a second opinion

simulacrum (Sep 16 2019 at 21:57, on Zulip):

cc @eddyb

ecstatic-morse (Sep 16 2019 at 22:58, on Zulip):

Just a heads up, #64470 contains a more generic dataflow analysis engine needed for const qualification, and I'm planning to make the current one a special-case of that one. I should be able to stay compatible with consumers of the old API though. Also, can we keep state_for_location private? It makes it too easy to accidentally go quadratic IMO.

ecstatic-morse (Sep 17 2019 at 01:43, on Zulip):

See #64532

gnzlbg (Sep 22 2019 at 08:24, on Zulip):

@ecstatic-morse FWIW you don't need to stay compatible with the old API

gnzlbg (Sep 22 2019 at 08:24, on Zulip):

if you change the API, clippy fails to compile, and they'll have to upgrade to the new API

gnzlbg (Sep 22 2019 at 08:26, on Zulip):

right now clippy is not using it, so maybe consider these type of users for the new API ? (I find the new one much better than the old one, mainly because it is much better documented)

Last update: Sep 18 2020 at 21:00UTC