Re: master fails to build/check

[Ladislav Michl]

On Fri, Jan 31, 2020 at 01:54:35PM -0800, Gary E. Miller wrote:
> Yo Ladislav!
> 
> On Fri, 31 Jan 2020 17:01:54 +0100
> Ladislav Michl <address@hidden> wrote:
> 
> > > - if SConstruct is using sysroot as all other packages, passing
> > > --sysroot to toolchain is just wrong.  
> >  - sysroot is indeed toolchain --sysroot=
> 
> Good.  We agree that sysroot is sysroot, as defined by GNU gcc and ld.

What a pleasant way to explain sysroot is sysroot, as defined by
GNU gcc and ld. I would say this deserves short notice somewhere,
as it might easily happen someone else would resist to believe
there is build system out there really setting toolchain sysroot.

> > and all libraries gpsd
> > depends on are discoverable using pkg-config.
> 
> Not even close.  Forget you ever thought that.

I will not, sorry.

> > Then most people do not
> > need to use sysroot at all,
> 
> Good thing most people do not use it then.
> 
> > but then we need something like this:
> 
> Lost me.  Instead of trying to fix what you admit you do not understand,
> can you explain what it is that you need?
> 
> I'm gonna guess that you do not want the host systemd tickled when you
> cross compile?

Obviously.

>  Or??

Or there are people out there wanting tickle host systemd while
cross-compiling?

> 
> > and then this code is still questionable:
> 
> Oh, good thing you say that, just what I was thinkgin too.
> 
> > # Honor the specified installation prefix in link paths.
> > if env["sysroot"]:
> >     env.Prepend(LIBPATH=[env["sysroot"] + installdir('libdir',
> >                 add_destdir=False)])
> 
> Lost me.  Is that a patch?  Current code?  What?

This is current SConstruct code.

> > Knowing what sysroot is will help fixing python modules cross-build.
> 
> Easy.  sysroot is the thing passed to gcc and ld as the value of --sysroot=

Thank you for confirming this.

> The place the cross-dev stuff is.

Just toolchains headers and libraries, it is insane to put anything else
here ;-)