Oddities with the dev86 package

Oddities with the dev86 package

[Christopher Baines]

[-- Attachment #1: Type: text/plain, Size: 1435 bytes --]

Hey,

I noticed that the dev86 package was a bit odd from looking at the
package reproducibility page in the Guix Data Service, it was appearing
in the "Matching" section for all architectures, which seemed a bit odd
[1].

1: http://data.guix.gnu.org/revision/3c6aca4232d1a3638ec962bc7afe9121626c43ec/package-reproducibility

Playing with this package on the command line, it seems that whatever
system you ask Guix to generate a derivation for, it generates the same
derivation.

→ guix build -d --system=i686-linux dev86
/gnu/store/wmzryaqgxxank355d9c1i7ikz813j7qd-dev86-0.16.21.drv

→ guix build -d --system=x86_64-linux dev86
/gnu/store/wmzryaqgxxank355d9c1i7ikz813j7qd-dev86-0.16.21.drv

→ guix build -d --system=aarch64-linux dev86
/gnu/store/wmzryaqgxxank355d9c1i7ikz813j7qd-dev86-0.16.21.drv


This is particularly odd, as the package declares it only supports
i686-linux and x86_64-linux, so I wouldn't expect to be able to compute
a derivation for aarch64-linux.

I'm guessing this behaviour has something to do with the #:system
argument to the build system?

From looking at the package definition myself, I'd expect at least that
generating a aarch64-linux derivation would fail. I'm not sure what
should happen in the x86_64-linux case, I guess you can infer some
information from asking for an x86_64-linux derivation, and then getting
back a i686-linux one.

Thanks,

Chris

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 962 bytes --]