Re: $ORIGIN in RPATH suppression

[pin <voidpin%protonmail.com@localhost>]

On Thursday, January 2nd, 2025 at 5:57 PM, Greg Troxel <gdt%lexort.com@localhost> wrote:

> pin voidpin%protonmail.com@localhost writes:
> 
> > > That's not really a solution, as anyone should be able to, using code in
> > > pkgsrc, build bootstraps.
> > 
> > I don't think this was ever true for Rust, as it all starts with an
> > amd64 compiler provided by upstream Rust from which all others are
> > produced or, am I wrong?
> 
> 
> What I meant is that while he@ does a ton of work (for which I am
> grateful), we should not have magic people who can do things that others
> can't. Whatever it takes to build things that pkgsrc needs -- even if
> that that starts with upstream binaries at the moment -- should be
> doable from bits that are checked in, by anyone. Just like anyone can
> run build.sh and set up pkgsrc bulk builds.
> 
> So separately from how the world ought to be (that rust upstream
> inexplicably doesn't value), I mean that anyone checking out pkgsrc
> should be able to do anything he@ can do, so "he@ carries patches that
> aren't committed" isn't a reasonable approach.

Fair enough, thanks for explaining, that didn't really came through in the initial short version.

But, how do we go from here?
Right now we have 1.81, although I'm confident that 1.82 could in principle be merged.
What to do with 1.83? And in 7 days 1.84?

I've commited nearly 1000 packages during 2024, most of it Rust things + LXQt (twice). I'm already holding two or three updates that require either 1.82 or, 1.83 to build.

I'm not trying to say anyone is wrong but, rather than focusing on the problem, I would prefer to find a solution, even if that's not the perfect one for everybody.

I do hope from some "magic" here.