Guile 2.0.14: check fails on i686

  • Done
  • quality assurance status badge
Details
3 participants
  • Denis 'GNUtoo' Carikli
  • Ludovic Courtès
  • Maxime Devos
Owner
unassigned
Submitted by
Denis 'GNUtoo' Carikli
Severity
important
D
D
Denis 'GNUtoo' Carikli wrote on 23 Jan 2022 01:33
(address . bug-guix@gnu.org)
20220123013327.3d412a7e@primary_laptop
Hi,

With Guix system i686, at the following commit:
Toggle quote (1 lines)
> cadcbbaf65 gnu: r-bigmelon: Add r-knitr.
guile 2.0.14 fails tests.

I've attached the log.

I'm unsure what to do here as it's Guile 2.0.14 and on Guile website I
only see Guile 3.0.7 and guile 2.2.7[1], and Guile 2.0.14 s from
2017[2].

References:
-----------

Denis.
-----BEGIN PGP SIGNATURE-----
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=PsPD
-----END PGP SIGNATURE-----


M
M
Maxime Devos wrote on 23 Jan 2022 11:47
840f7adc79c46bd7e0a16b368fd1d3273e3d04d3.camel@telenet.be
Denis 'GNUtoo' Carikli schreef op zo 23-01-2022 om 01:33 [+0100]:
Toggle quote (12 lines)
> Hi,
>
> With Guix system i686, at the following commit:
> > cadcbbaf65 gnu: r-bigmelon: Add r-knitr.
> guile 2.0.14 fails tests.
>
> I've attached the log.
>
> I'm unsure what to do here as it's Guile 2.0.14 and on Guile website I
> only see Guile 3.0.7 and guile 2.2.7[1], and Guile 2.0.14 s from
> 2017[2].

FAIL: numbers.test: Number-theoretic division: euclidean/: mixed types:
(130.0 10/7) [other errors]

See https://issues.guix.gnu.org/49368. That's for Guile 3.0.7, but
it seems like the same issue exists for Guile 2.0.14.

It seems like the fix needs to be moved to guile-2.0. Some care might
be required to avoid rebuilds.

Greetings,
Maxime.
-----BEGIN PGP SIGNATURE-----

iI0EABYKADUWIQTB8z7iDFKP233XAR9J4+4iGRcl7gUCYe0yThccbWF4aW1lZGV2
b3NAdGVsZW5ldC5iZQAKCRBJ4+4iGRcl7lCVAP9gFFpfjlb3irg5nOnmtmHyjt3b
tbyTXTl25dSGjICCsAEA8Z78OOPiMFgAxgId/aUul1iP62/SZo9LeZeb8TL/KwE=
=hdFJ
-----END PGP SIGNATURE-----


M
M
Maxime Devos wrote on 23 Jan 2022 11:58
08ab95f003fa906efa63b42e1add869e9390ae34.camel@telenet.be
Denis 'GNUtoo' Carikli schreef op zo 23-01-2022 om 01:33 [+0100]:
Toggle quote (3 lines)
> > cadcbbaf65 gnu: r-bigmelon: Add r-knitr.
> guile 2.0.14 fails tests.

According to "guix refresh -l", ‘only’ 83 packages depend on
guile@2.0.14:

guile2.0-shepherd@0.8.1 libchop@0.5.2 geda-gaf@1.10.2 gnubik@2.4.3
guile-gnunet@0.0-1.d12167a guile2.0-gcrypt@0.3.0 guile2.0-redis@2.1.1
guildhall@0-1.2fe2cc539 guile2.0-pg@0.49 guile2.0-filesystem@0.1.0
guile2.0-bash@0.1.6-0.1eabc56 guile2.0-irregex@0.9.6 guile2.0-
commonmark@0.1.2 guile2.0-haunt@0.2.5 guile2.0-lib@0.2.7 guile2.0-
minikanren@20150424.e844d85 guile2.0-sqlite3@0.1.2 guile2.0-gdbm-
ffi@20120209.fa1d5b6 guile2.0-git@0.5.2 prusa-slicer@2.4.0 mia@2.4.6 r-
mlr3ordinal@0.1.0-9000-0.9febac9 r-actcd@1.2-0 r-abn@2.5-0 r-
scdd@1.18.0 r-glmmtmb@1.1.2.3 r-weights@1.0.4 r-semplot@1.1.2 r-
altmeta@3.3 r-blme@1.0-5 r-projpred@2.0.2 r-tidyposterior@0.1.0 r-
sjplot@2.8.10 r-mitml@0.4-3 r-variancepartition@1.24.0 r-
rbamtools@2.16.17 r-r2glmm@0.1.2 r-bionero@1.2.0 r-bioconcotk@1.14.0 r-
cytonorm@0.0.7-1.e4b9d34 pigx@0.0.3 r-shapforxgboost@0.1.1 r-
survminer@0.4.9 r-ez@4.4-0 r-catalyst@1.18.0 r-destiny@3.8.0 r-
analytics@3.0 r-spectre@0.5.5-1.f6648ab r-radiant-data@1.4.1 r-
effects@4.2-0 freetalk@4.1 denemo@2.5.0 lsh@2.1 guile2.0-ssh@0.13.1
guile2.0-gnutls@3.7.2 zile-on-guile@2.6.2-0.fd09781

The R packages seem to require precisely guile@2.0.14 through 'nlopt'.
Maybe nlopt works with guile@2.2, maybe it can easily be modified to
support guile@2.2 (or even guile@3.O). We could also disable the
guile bindings in 'nlop' when compiling for target-x86-32?.

Greetings,
Maxime.
-----BEGIN PGP SIGNATURE-----

iI0EABYKADUWIQTB8z7iDFKP233XAR9J4+4iGRcl7gUCYe003hccbWF4aW1lZGV2
b3NAdGVsZW5ldC5iZQAKCRBJ4+4iGRcl7nl+AQDx+iyIDP1qpYfFDlLBgkJJ0pgn
lkGXHxT+aMzY9gODrwEA3f99DUZBbh/op/qj/fwoU/wpvE89TAqePhlPhyVlaw4=
=KK+O
-----END PGP SIGNATURE-----


L
L
Ludovic Courtès wrote on 24 Jan 2022 15:22
control message for bug #53462
(address . control@debbugs.gnu.org)
8735ldusbk.fsf@gnu.org
severity 53462 important
quit
L
L
Ludovic Courtès wrote on 24 Jan 2022 15:42
Re: bug#53462: Guile 2.0.14: check fails on i686
(name . Maxime Devos)(address . maximedevos@telenet.be)
87h79ttcu0.fsf@gnu.org
Hi,

Maxime Devos <maximedevos@telenet.be> skribis:

Toggle quote (9 lines)
> FAIL: numbers.test: Number-theoretic division: euclidean/: mixed types:
> (130.0 10/7) [other errors]
>
> See <https://issues.guix.gnu.org/49368>. That's for Guile 3.0.7, but
> it seems like the same issue exists for Guile 2.0.14.
>
> It seems like the fix needs to be moved to guile-2.0. Some care might
> be required to avoid rebuilds.

Indeed. Done in c9627323eca0fb1e0b0a0680049ea0f9f20def5e.

Thanks!

Ludo’.
Closed
?