Re: git: 06601897e5cd - main - framework: reintroduce the feature enabling code

From: Mathieu Arnold <mat_at_freebsd.org>
Date: Sat, 13 Jul 2024 14:30:31 UTC
On Sat, Jul 13, 2024 at 02:04:48PM GMT, Daniel Engberg wrote:
> On 2024-07-13T08:47:47.000+02:00, Mathieu Arnold <mat@freebsd.org>
> wrote:
> 
> > On Sat, Jul 13, 2024 at 06:11:04AM GMT, Daniel Engberg wrote:
> >>  Hi,
> >>  
> >>   This changes so LTO option is no longer applied to Rust (cargo)
> >>  ports
> >>  
> >>   BY DEFAULT causing a regresssion, please fix.
> > 
> > As it has been three months, nobody complained something was broken
> > so,
> > 
> > I don't think anything is actually broken.
> > 
> > LTO as are a few other features like SSP are user facing features,
> > not a
> > 
> > porters facing options, it means, it's up to the person doing the
> > 
> > building to choose wether to enable it or not, it is **not** up to
> > the
> > 
> > person porting the software to forcefully enable it.
> > 
> > -- 
> > 
> > Mathieu Arnold
> 
> Hi,
> 
> Likely because this pretty much silently went by because it was posted
> on Phab and you only CCed bapt. It's been enabled since Jan 2023
> (https://cgit.freebsd.org/ports/commit/Mk/Uses/cargo.mk?id=967022fd812cf67dec264ee4e53bd016b69e7a2b)
> and tested/discussed here https://reviews.freebsd.org/D36736 before
> being enabled/committed. I noticed it now while updating a Rust
> (cargo-based) port.

Mmmm, yes, I know about that, and I agree, this commits reverts this
behavior.

Because choosing to build with or without LTO is a user facing feature,
not a porter facing feature, so, it has to be set by people building the
things, not by the framework or a port.

-- 
Mathieu Arnold