Stream: t-compiler

Topic: bootstrapping cfg


gnzlbg (Sep 22 2019 at 10:39, on Zulip):

When adding a new intrinsic to libcore, one used to be able to write #[cfg(not(stage0))] to avoid crashing the bootstrap compiler

gnzlbg (Sep 22 2019 at 10:39, on Zulip):

I've tried that and #[cfg(not(bootstrap))] and neither work

gnzlbg (Sep 22 2019 at 10:39, on Zulip):

did the cfg that gets passed change ?

oli (Sep 22 2019 at 12:12, on Zulip):

the latter should work, I've used it successfully before

oli (Sep 22 2019 at 12:12, on Zulip):

maybe we're in boostrap mode again?

oli (Sep 22 2019 at 12:13, on Zulip):

something about SIMD in libcore

simulacrum (Sep 22 2019 at 12:27, on Zulip):

@gnzlbg I think .. let me check, if Alex's PR hasn't landed yet you need boostrap_please_ignore_stdarch or os

simulacrum (Sep 22 2019 at 12:27, on Zulip):

Yeah, @gnzlbg use cfg(boostrap_stdarch_ignore_this)

simulacrum (Sep 22 2019 at 12:27, on Zulip):

(yes, typo from bootstrap is "intentional")

gnzlbg (Sep 22 2019 at 12:42, on Zulip):

that worked :D

gnzlbg (Sep 22 2019 at 12:43, on Zulip):

I found that below in the file

gnzlbg (Sep 22 2019 at 12:43, on Zulip):

but i couldn't grep for it

gnzlbg (Sep 22 2019 at 12:43, on Zulip):

because i was obviously greeping for bootstrap and not boostrap

gnzlbg (Sep 22 2019 at 12:44, on Zulip):

what revelead it was the src/tool/.../bootstrap.rs file, where i could find both --cfgs being passed

gnzlbg (Sep 22 2019 at 12:44, on Zulip):

i was only briefly confused by the typo though

Jake Goulding (Sep 24 2019 at 16:59, on Zulip):

semi-related https://twitter.com/boostrapghost

Last update: Nov 22 2019 at 04:30UTC