Re: building certbot on Sparc Solaris10...

["K. Schreiner" <ks%ub.uni-mainz.de@localhost>]

On Wed, Dec 28, 2022 at 08:48:00AM +0000, nia wrote:
> On Tue, Dec 27, 2022 at 09:27:39PM +0100, K. Schreiner wrote:
> > Thanks for the quick answer, but - nothing changed, sorry.
> > 
> > Typing "bmake" in pkgsrc/security/py-cryptography yields the same output as with the origial pkgsrc/lang/rust/platform.mk.
> > 
> > 
> > What I thought would be possible is to compile pkgsrc/security/py-cryptography w/o rust using C, depending on rust beeing available or not.
> > (If using C for py-cryptography is still possible at all, as it once was.)
> > 
> > FWIW: I tried to use py-cryptography from pkgsrc_2022Q2, but that led to runtime errors.
> > 
> > For the time beeing I'll use the last compileable certbot from pkgsrc_2022Q2... Or switch to acmesh, f.ex. - anything not using rust will do. ;-)
> > 
> > 
> > Thanks again, Kurt
> 
> That's not how it works.
> 
> security/py-cryptograpy is the rust version. Typing bmake inside
> there will try to build the rust version.
> 
> security/py27-cryptograpy is the old non-rust version.
> 
> The purpose of setting rust as unavailable on Solaris/sparc64 is
> to make certbot pick security/py27-cryptograpy as a dependency
> instead.

looking in the wrong place:

the culprit is lang/rust/platform.mk stating:

...
Result[] of :U is "SunOS-*-sparc64"
Pattern[MACHINE_PLATFORM] for [SunOS-5.10-sparc64] is [SunOS-*-sparc64]
VarMatch [SunOS-5.10-sparc64] [SunOS-*-sparc64]
Result[MACHINE_PLATFORM] of :M is "SunOS-5.10-sparc64"
Global:PLATFORM_SUPPORTS_RUST = yes
Applying[MACHINE_PLATFORM] :M to "SunOS-5.10-sparc64"
Applying[] :U to ""
...

but PLATFORM_SUPPORTS_RUST should be false for SunOS-5.10-sparc64.


HTH, Kurt