Stream: t-compiler

Topic: #55396 linker flavour inference


davidtwco (Nov 29 2018 at 15:14, on Zulip):

Spawning off from weekly meeting.

davidtwco (Nov 29 2018 at 15:17, on Zulip):

cc @nagisa

davidtwco (Nov 29 2018 at 15:21, on Zulip):

It seems like the first proposed change to the inference propsed (only stripping the extension when it is .exe) seems trivial enough. I'm not too familiar with what would be required to stablize -C linker-flavour though.

nagisa (Nov 29 2018 at 15:21, on Zulip):

Stabilizing -Clinker-flavour would involve making a patch and getting it go through a FCP

nagisa (Nov 29 2018 at 15:22, on Zulip):

which is also quite easy (usually involves removing code, not adding it)

davidtwco (Nov 29 2018 at 15:23, on Zulip):

Is this guide up-to-date? https://forge.rust-lang.org/stabilization-guide.html

nagisa (Nov 29 2018 at 15:24, on Zulip):

Yeah.

davidtwco (Nov 29 2018 at 15:24, on Zulip):

Alright, thanks, I should be good to get started on this.

nikomatsakis (Nov 29 2018 at 15:27, on Zulip):

nice

nagisa (Nov 29 2018 at 17:59, on Zulip):

Ah, also @davidtwco you’ll want to keep the two PRs (one that only strips the exe extension and the stabilisation one) separate

davidtwco (Nov 29 2018 at 17:59, on Zulip):

Yeah, I figured that'd be the case :thumbs_up:

davidtwco (Nov 29 2018 at 19:29, on Zulip):

@nagisa are there tests for the linker stuff anywhere?

nagisa (Nov 29 2018 at 19:37, on Zulip):

There should be at least a test for the -Z flag.

nagisa (Nov 29 2018 at 19:39, on Zulip):

There does not appear to be a test for -Clinker operation though

davidtwco (Nov 29 2018 at 19:39, on Zulip):

I can't find any for the linked flavor inference - PR that added it doesn't have any.

nagisa (Nov 29 2018 at 19:39, on Zulip):

and it is super platform-specific anyway…

nagisa (Nov 29 2018 at 19:39, on Zulip):

I’m not sure how you'd even check what linker rustc picks with given flags…

davidtwco (Nov 29 2018 at 19:40, on Zulip):

Fair enough, I guess I'll just open a PR with what I've got then.

davidtwco (Nov 29 2018 at 19:41, on Zulip):

It's really difficult to not spell flavour with the 'u'.

nagisa (Nov 29 2018 at 19:43, on Zulip):

I’ve just given up on not spelling the non-american english way

davidtwco (Nov 29 2018 at 19:44, on Zulip):

I've just given up on not spelling the non-american english way

?

nagisa (Nov 29 2018 at 19:45, on Zulip):

the double-negative was intended :slight_smile: I just spell the british way instead.

davidtwco (Nov 29 2018 at 19:45, on Zulip):

Ah, I wasn't sure if I was parsing that wrong or not.

davidtwco (Nov 29 2018 at 19:48, on Zulip):

@nagisa submitted #56349 with the first change

davidtwco (Nov 29 2018 at 21:02, on Zulip):

@nagisa submitted #56351 with the stabilization.

mw (Nov 30 2018 at 10:57, on Zulip):

With -Clinker-flavour stabilized, can we now default to LLD on major platforms?

nagisa (Nov 30 2018 at 11:19, on Zulip):

@mw I see no problem changing to lld by default even without stabilising that

nagisa (Nov 30 2018 at 11:19, on Zulip):

as -Clinker will infer the correct linker flavour

Jake Goulding (Dec 01 2018 at 16:27, on Zulip):

Did the British English version get stabilized? I thought that the compiler options were American English.

davidtwco (Dec 01 2018 at 16:31, on Zulip):

The option was always in American English IIRC. Only place I ever used British English was in commit messages (before I caught myself and changed it).

nagisa (Dec 01 2018 at 16:40, on Zulip):

Yes, we use AmE everywhere in the compiler. That may change once we get localisation of the diagnostics and error code descriptions

nagisa (Dec 01 2018 at 16:40, on Zulip):

but till then…

Jake Goulding (Dec 01 2018 at 16:42, on Zulip):

I was just seeing :

With -Clinker-flavour stabilized,

nagisa (Dec 01 2018 at 16:43, on Zulip):

That’s just me being silly and typing too fast to correct myself later

Jake Goulding (Dec 01 2018 at 16:43, on Zulip):

And mw as well ;-)

Last update: Nov 16 2019 at 01:20UTC