Closed strogdon closed 2 years ago
An oversight. There are a few ebuilds that may need updating. The latest sagetex is also capped at python 3.9.
Ok let's see if it fixes your issue.
It fixed ExportSageNB
but you are correct in that sagetex
now prevents using only python3.10
. I changed sagetex
in my overlay to be PYTHON_COMPAT=( python3_{8..10} )
and that seems to be the last bit needed to build only against python3.10
.
I have pushed that too https://github.com/cschwan/sage-on-gentoo/commit/739b16bd623ec4d64e97d1c35987a4b14f1edcf4 how come you didn't get it?
I'm wrong. There is something else. I can't just do
=sci-mathematics/sage-9999 -python_targets_python3_9
with PYTHON_TARGETS="python3_9 python3_10"
Is this normal?
sage_docbuild
requires sage
with matching pythons as PDEPEND.
I think it would be best to try PYTHON_TARGETS=python3_10 emerge -1v @sagemath
Relative to commits. I believe you fixed sagetex-3.6
but not sagetex-3.5-r1
which is required for sage-9999
.
Darn, yes. Another oversight. I don't believe there is any reason for that.
OK, sagetex 3.5 PYTHON_COMPAT bumped. I'll have to check the reason the update to 3.6 is stalled in vanilla.
Yes, even
=sci-mathematics/sage-9999 -python_targets_python3_9
=sci-mathematics/sage_docbuild-9999 -python_targets_python3_9
will not work with PYTHON_TARGETS="python3_9 python3_10"
in make.conf
. It is simply ignored. PYTHON_TARGETS
must be set in the emerge
.
508 packages to build. I hope my hard drive is up for the task.
I don't have PYTHON_TARGETS
in make.conf
anymore. I have /etc/portage/package.use/python
which looks like
*/* PYTHON_TARGETS: -* python3_9 python3_10
which suggests how to structure package.use
files for selective python. The alternative would be to use env
and package.env
files.
Thanks for the pointers. This seems to work
=sci-mathematics/sage-9999 PYTHON_TARGETS: -* python3_10
=sci-mathematics/sage_docbuild-9999 PYTHON_TARGETS: -* python3_10
=sci-mathematics/sage_setup-9999 PYTHON_TARGETS: -* python3_10
Nice to know it does.
sbcl
will not build on my new Prefix (https://bugs.gentoo.org/837746). clisp
does build. Any downside of using clisp
instead of sbcl
in building maxima
for Sage?
None. So long as ecls
works there are no issues. My own maxima install is ecls
only.
OK, I try the ecls
only approach. I'm not sure what's up with sbcl
.
Always been a funny one sbcl
. You need sbcl
to bootstrap sbcl
. There is probably an architecture confusion on prefix.
I noticed that.
Sage finally built on the Prefix using only python3.10
. To document things, I had the following build issues with (solutions)
sbcl
failed to build (built maxima
with USE=-sbcl)sage-data-conway_polynomials
was installed but nothing was really installed (reinstalled the package) - the indicater of the problem was that sage- doc
failed because conway_polynomials.p
could not be foundgiac
failed the install phase with /storage/strogdon/gentoo-rap/usr/bin/install: cannot create regular file '/storage/strogdon/gentoo-rap/var/tmp/portage/sci-mathematics/giac-1.7.0.13-r1/image/storage/strogdon/gentoo-rap/usr/share/giac/doc/es/xcasex': File exists
. I believe I've seen this previously. (I restarted the install using portage folder contents with ebuild giac ... install
and then ebuild giac ... qmerge
)The main issue - to build Sage with just python3.10
has succeeded. Thanks for the assistance.
My Prefix somehow got corrupted - perhaps a hard drive issue. I reinstalled a new Prefix and so far OK. To save cpu cycles I tried to install
sage-9999
withoutpython3.9
and I gotI removed
python3.9
fromPYTHON_TARGETS
and discovered thatExportSageNB-3.3
needs that python. I believe vanillasagenb_export
builds againstpython3.10
. Is there any reason that shouldn't work with s-o-g. I'm assuming that thepython3.9
requirement to buildsage-9999
is because ofExportSageNB-3.3
.