Stream: t-compiler

Topic: PlaceContext for AscribeUserType


RalfJ (Oct 23 2018 at 15:25, on Zulip):

The default mir visitor impl generated by make_mir_visitor! classifies places used in a AscribeUserType as PlaceContext::Validate. that's clearly wrong, that PlaceContext is for miri's validation statements -- I noticed this now because I am changing Validate to Retag. So AscribeUserType gets its own variant in PlaceContext, but what I am not sure about is whether this should be classified as mutating use, nonmutating use or neither. Validate was neither but that seems wrong for an AscribeUserType?

nikomatsakis (Oct 23 2018 at 16:21, on Zulip):

@RalfJ it is neither

nikomatsakis (Oct 23 2018 at 16:21, on Zulip):

I remember now that I wanted to change that context

nikomatsakis (Oct 23 2018 at 16:21, on Zulip):

ascribe-user-type has no effect except to inform the NLL type-checker

RalfJ (Oct 23 2018 at 21:57, on Zulip):

kk

Last update: Nov 16 2019 at 01:35UTC