Stream: t-compiler

Topic: design space for structural_match and friends


pnkfelix (Oct 17 2019 at 13:10, on Zulip):

At this point I've explored the design space a fair amount and think I could present semi-coherently on-the-fly if needed

centril (Oct 17 2019 at 13:10, on Zulip):

:slight_smile:

pnkfelix (Oct 17 2019 at 13:10, on Zulip):

but I agree that I want to take a conservative approach initially

pnkfelix (Oct 17 2019 at 13:10, on Zulip):

which is actually good

pnkfelix (Oct 17 2019 at 13:11, on Zulip):

because my understanding of "conservative approach" is actually a version I already got working, and then set aside to experiment with miri values

centril (Oct 17 2019 at 13:12, on Zulip):

(to elaborate: what I mean by "conservative" is "accept less code" ^^)

pnkfelix (Oct 17 2019 at 13:12, on Zulip):

(and now discussions with @nikomatsakis have convinced me that while miri values could be great to use in the long term, they could also present a stability hazard for const fn that users will not expect)

pnkfelix (Oct 17 2019 at 13:12, on Zulip):

anyway lets maybe table this or at least fork it off to a separate topic

centril (Oct 17 2019 at 13:12, on Zulip):

agree

pnkfelix (Oct 17 2019 at 13:13, on Zulip):

in fact I'll fork it now

Last update: Nov 16 2019 at 01:40UTC