Sage 9.8.rc1 builds, but reports an empty list of packages failed.

66 views
Skip to first unread message

David Kirkby

unread,
Feb 10, 2023, 3:17:38 PM2/10/23
to sage-support
I got this rather weird error on Ubuntu 20.04. It says the following packages failed to build, but does not say what they were! The hardware is a Dell 7920 with 2 x 26 core CPUs and 384 GB RAM.

I put config.log here

and a compressed version of install.log here


make[3]: *** [Makefile:3262: /home/drkirkby/sage/local/var/lib/sage/installed/sagemath_doc_html-none] Error 2
make[2]: *** [Makefile:2766: all-start] Error 2
make[2]: Leaving directory '/home/drkirkby/sage/build/make'

real    47m17.658s
user    220m4.574s
sys    14m6.540s
****************install.log.bz2***********************************************
Error building Sage.

The following package(s) may have failed to build (not necessarily
during this run of 'make all-start'):

It is safe to delete any log files and build directories, but they
contain information that is helpful for debugging build problems.
WARNING: If you now run 'make' again, the build directory of the
same version of the package will, by default, be deleted. Set the
environment variable SAGE_KEEP_BUILT_SPKGS=yes to prevent this.

make[1]: *** [Makefile:40: all-start] Error 1
make[1]: Leaving directory '/home/drkirkby/sage'
make: *** [Makefile:13: all] Error 2

Trying to run the tests, it reports the following, which is not in the log file uploaded.

real    0m55.972s
user    8m4.622s
sys    0m47.623s
***************************************************************
Error building Sage.

The following package(s) may have failed to build (not necessarily
during this run of 'make all-start'):

* package:         sagemath_doc_html-none
  last build time: Feb 10 20:12
  log file:        /home/drkirkby/sage/logs/pkgs/sagemath_doc_html-none.log

It is safe to delete any log files and build directories, but they
contain information that is helpful for debugging build problems.
WARNING: If you now run 'make' again, the build directory of the
same version of the package will, by default, be deleted. Set the
environment variable SAGE_KEEP_BUILT_SPKGS=yes to prevent this.

make[1]: *** [Makefile:40: all-start] Error 1
make[1]: Leaving directory '/home/drkirkby/sage'
make: *** [Makefile:13: all] Error 2



David Kirkby

unread,
Feb 10, 2023, 4:10:47 PM2/10/23
to sage-support
It looks like sagemath_doc_html-none.is failing to build - full log of that package only at


This is the end of the logfile.

Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/modfrm
[combinat ] no targets are out of date.
[combinat ] The inventory files are in ../../local/share/doc/sage/inventory/en/reference/combinat.
Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/combinat
Error in glue(.Internal(R.home()), "library", "base", "R", "base", sep = .Platform$file.sep) :
  4 arguments passed to .Internal(paste) which requires 3
Error: could not find function "attach"
Error: object '.ArgsEnv' not found
Fatal error: unable to initialize the JIT

make[4]: *** [Makefile:28: doc-inventory--reference-interfaces] Error 2
make[3]: *** [Makefile:45: doc-inventory-reference] Error 2
(base) drkirkby@canary:~/sage$ scp /home/drkirkby/sage/logs/pkgs/sagemath_doc_html-none.log k...@kirkbymicrowave.co.uk:
sagemath_doc_html-none.log                                                                                                 100%  162KB   2.5MB/s   00:00    
(base) drkirkby@canary:~/sage$

Dima Pasechnik

unread,
Feb 11, 2023, 8:39:52 AM2/11/23
to sage-s...@googlegroups.com
On Fri, Feb 10, 2023 at 9:10 PM David Kirkby <drki...@gmail.com> wrote:
It looks like sagemath_doc_html-none.is failing to build - full log of that package only at


This is the end of the logfile.

Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/modfrm
[combinat ] no targets are out of date.
[combinat ] The inventory files are in ../../local/share/doc/sage/inventory/en/reference/combinat.
Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/combinat
Error in glue(.Internal(R.home()), "library", "base", "R", "base", sep = .Platform$file.sep) :
  4 arguments passed to .Internal(paste) which requires 3
Error: could not find function "attach"
Error: object '.ArgsEnv' not found
Fatal error: unable to initialize the JIT


I.e. apparently there is something amiss with R/rpy2 in your case.

Could you provide the top-level config.log
along with logs/pkgs/rpy*.log

We've switched to using system-wide install of R by default some time ago, perhaps your R is too new for rpy2 we ship ?

Dima

 

--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-support/03f57707-bed0-4214-8e2e-59a8ac500bf9n%40googlegroups.com.

David Kirkby

unread,
Feb 11, 2023, 3:33:04 PM2/11/23
to sage-support
On Saturday, 11 February 2023 at 13:39:52 UTC Dima Pasechnik wrote:
On Fri, Feb 10, 2023 at 9:10 PM David Kirkby <drki...@gmail.com> wrote:
It looks like sagemath_doc_html-none.is failing to build - full log of that package only at


This is the end of the logfile.

Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/modfrm
[combinat ] no targets are out of date.
[combinat ] The inventory files are in ../../local/share/doc/sage/inventory/en/reference/combinat.
Build finished. The built documents can be found in /home/drkirkby/sage/local/share/doc/sage/inventory/en/reference/combinat
Error in glue(.Internal(R.home()), "library", "base", "R", "base", sep = .Platform$file.sep) :
  4 arguments passed to .Internal(paste) which requires 3
Error: could not find function "attach"
Error: object '.ArgsEnv' not found
Fatal error: unable to initialize the JIT


I.e. apparently there is something amiss with R/rpy2 in your case.

Could you provide the top-level config.log
along with logs/pkgs/rpy*.log

We've switched to using system-wide install of R by default some time ago, perhaps your R is too new for rpy2 we ship ?

Dima

Hi Dima,

This is my R version
(base) drkirkby@canary:~/sage/logs/pkgs$ R --version
R version 4.2.2 (2022-10-31) -- "Innocent and Trusting"

It's not an official Ubuntu version, which would be in /usr/bin

(base) drkirkby@canary:~$ command -v R
/home/drkirkby/miniforge3/bin/R

I did actually provide a link to the config.log before.
as well as a compressed version of install.log

Note I run make -j, so the log might look a bit messy, as it would have used as many processes as it could to hopefully reduce the compilation time. 

I've attached rpy2-3.4.5.log

I did not actually intend to build a release candidate, but that's what I got following the instructions. Is the norm now to put issues with release candidates on sage-devel in future? I don''t know what the norm is now.

As you may recall, my only involvement in Sage development was porting it to Solaris. I'm an engineer, not a mathematician, and consider myself relatively weak at maths. I'm now thinking of doing an Open University (OU) maths degree
just for something to do. Hence I thought I'd take a look at Sage again. Unfortunately, at least on the statistics modules I would have to do,  the OU use Minitab. I'd much rather use the open-source R, but would not have any say in the matter. I guess they think the can teach statistics quicker with Minitab, as one does not have to do any programming.

Dave

rpy2-3.4.5.log

Dima Pasechnik

unread,
Feb 11, 2023, 4:06:10 PM2/11/23
to sage-s...@googlegroups.com
Oh, you build in Conda...
This might be a bit of a problem - it's not too well-tested for sure.

 

I did actually provide a link to the config.log before.
as well as a compressed version of install.log

Note I run make -j, so the log might look a bit messy, as it would have used as many processes as it could to hopefully reduce the compilation time. 

I've attached rpy2-3.4.5.log

I did not actually intend to build a release candidate, but that's what I got following the instructions. Is the norm now to put issues with release candidates on sage-devel in future? I don''t know what the norm is now.

As you may recall, my only involvement in Sage development was porting it to Solaris. I'm an engineer, not a mathematician, and consider myself relatively weak at maths. I'm now thinking of doing an Open University (OU) maths degree
just for something to do. Hence I thought I'd take a look at Sage again. Unfortunately, at least on the statistics modules I would have to do,  the OU use Minitab. I'd much rather use the open-source R, but would not have any say in the matter. I guess they think the can teach statistics quicker with Minitab, as one does not have to do any programming.

Dave

--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.

Dr. David Kirkby

unread,
Feb 11, 2023, 4:23:37 PM2/11/23
to sage-s...@googlegroups.com
On Sat, 11 Feb 2023 at 21:06, Dima Pasechnik <dim...@gmail.com> wrote:
On Sat, Feb 11, 2023 at 8:33 PM David Kirkby <drki...@gmail.com> wrote:
/home/drkirkby/miniforge3/bin/R

Oh, you build in Conda...
This might be a bit of a problem - it's not too well-tested for sure.

Yes, someone suggested that I believe, when I had another problem earlier. I basically followed the instructions there. It that's not tested well, I'll download
 
https://www.mirrorservice.org/sites/www.sagemath.org/devel/sage-9.8.rc1.tar.gz

and see what happens. I might as well do the release candidate - I'm not going to be doing anything serious with this now, so I don't mind the bugs. If there is a real problem building with the latest R, I guess that should be fixed before a new release.

I saw a comment from someone on one of the Sage mailing lists about the version of R being packaged with Sage is old.

Dave

Dima Pasechnik

unread,
Feb 11, 2023, 4:31:32 PM2/11/23
to sage-s...@googlegroups.com
On Sat, Feb 11, 2023 at 9:23 PM Dr. David Kirkby <drki...@gmail.com> wrote:
On Sat, 11 Feb 2023 at 21:06, Dima Pasechnik <dim...@gmail.com> wrote:
On Sat, Feb 11, 2023 at 8:33 PM David Kirkby <drki...@gmail.com> wrote:
/home/drkirkby/miniforge3/bin/R

Oh, you build in Conda...
This might be a bit of a problem - it's not too well-tested for sure.

Yes, someone suggested that I believe, when I had another problem earlier. I basically followed the instructions there. It that's not tested well, I'll download
 
https://www.mirrorservice.org/sites/www.sagemath.org/devel/sage-9.8.rc1.tar.gz


and see what happens. I might as well do the release candidate - I'm not going to be doing anything serious with this now, so I don't mind the bugs. If there is a real problem building with the latest R, I guess that should be fixed before a new release.

I saw a comment from someone on one of the Sage mailing lists about the version of R being packaged with Sage is old.

Dave

--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.

Dima Pasechnik

unread,
Feb 11, 2023, 4:57:25 PM2/11/23
to sage-s...@googlegroups.com
On Sat, Feb 11, 2023 at 9:23 PM Dr. David Kirkby <drki...@gmail.com> wrote:
we basically removed the option of building R from source - it's too much of an effort, and we don't have resources for it.

By the way, the preferred way to report Sage bugs is to open an issue on 
https://github.com/sagemath/sage/issues/ (yes, you'd need a GitHub account for it)


Your issue you have might be due to relatively old rpy2 we ship.
As far as I am concerned we can bump up rpy2 to a 3.5 version.


 


Dave

--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.

Dr. David Kirkby

unread,
Feb 11, 2023, 4:59:46 PM2/11/23
to sage-s...@googlegroups.com
On Sat, 11 Feb 2023 at 21:31, Dima Pasechnik <dim...@gmail.com> wrote:
Okay, I'm trying again.

It seems an unusual build process.
* There's a configure.ac, but no configure script.
* I typed make, which then builds a configure script, and exits with an error message, telling one to run configure.
* The configure script suggested I added some packages, which I did.
* I run "make distclean" with the intention of starting fresh. But noticed "make distclean" did not delete the configure script it had built earlier. So it seems "make distclean" is not really working in the usual way. 

I'm just running

make -j ptestlong

I don't know if that's still the way to run the tests. This is a reasonably quick machine (52 cores), so I should be able to get though the tests quickly if they run in parallel

After I'd written everything above, I just see your message about github. I'll do that if I get any more issues.

Dave


Dima Pasechnik

unread,
Feb 11, 2023, 5:11:09 PM2/11/23
to sage-s...@googlegroups.com
On Sat, Feb 11, 2023 at 9:59 PM Dr. David Kirkby <drki...@gmail.com> wrote:
On Sat, 11 Feb 2023 at 21:31, Dima Pasechnik <dim...@gmail.com> wrote:

Okay, I'm trying again.

It seems an unusual build process.
* There's a configure.ac, but no configure script.

you should have been running ./bootstrap-conda

but yes, the default tarball on https://github.com/sagemath/sage/releases is now
an unconfigured one - that's due to our transition from trac that things are a bit messy atm

 

* I typed make, which then builds a configure script, and exits with an error message, telling one to run configure.
* The configure script suggested I added some packages, which I did.
* I run "make distclean" with the intention of starting fresh. But noticed "make distclean" did not delete the configure script it had built earlier. So it seems "make distclean" is not really working in the usual way. 

I'm just running

make -j ptestlong

yes. Although -j without a parameter is a bit risky...
 

I don't know if that's still the way to run the tests. This is a reasonably quick machine (52 cores), so I should be able to get though the tests quickly if they run in parallel

After I'd written everything above, I just see your message about github. I'll do that if I get any more issues.

Dave


--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.

julian...@gmail.com

unread,
Feb 11, 2023, 5:22:37 PM2/11/23
to sage-support
Hi Dave,

On Saturday, February 11, 2023 at 10:33:04 PM UTC+2 David Kirkby wrote:
(base) drkirkby@canary:~$ command -v R
/home/drkirkby/miniforge3/bin/R

something went wrong in your conda setup I believe. With the instructions in the documentation, R should be installed in the sage-dev environment and not in the base environment. Probably that mix caused trouble in the process at some point.

julian

Dr. David Kirkby

unread,
Feb 11, 2023, 8:55:27 PM2/11/23
to sage-s...@googlegroups.com
Thank you. I was not too keen on that condra setup. I didn’t understand what it was doing. it changed my shell prompt, and increased the time for the prompt to appear, which I was not expecting. 

I have R in /usr/bin too. I dont know the version of that, and I am not near the computer. 

I started to build Sage from the source code, but stopped the process before it had finished. The computer is in my lab, which is part of the garage. I didn’t want to leave the computer on all night using power. It takes 200 W when idle. 😢😢😢😢😢😢 

I will restart later on. 

Dave

--
You received this message because you are subscribed to the Google Groups "sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.
--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drki...@kirkbymicrowave.co.uk
https://www.kirkbymicrowave.co.uk/
Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United Kingdom

Dr. David Kirkby

unread,
Feb 12, 2023, 1:22:58 PM2/12/23
to sage-s...@googlegroups.com
On Sat, 11 Feb 2023 at 22:11, Dima Pasechnik <dim...@gmail.com> wrote:


I'm just running

make -j ptestlong

yes. Although -j without a parameter is a bit risky...

Point taken Dima about using -j without a parameter. That could have been tricky if doctesting had used thousands of threads.😂😂

TLDR: I have sage  9.8 built and passed all tests with the latest R. The rpy in Sage is compatible with the latest R, and does not need updating just to work with the current version of R

There are a couple of things bugging me.

* Sage is only doctesting with 8 threads.
* A message "pytest is not installed in the venv, skip checking tests that rely on it" This is despite I have pytest - admittedly an old version. 
* A weird failure of 'make' when pytest was not installed.

This is what I did.

1) Deleted anything and everything related to Conda. The fact it had changed my prompt to have "(base)" in it, and the prompt seemed to take a long time to appear when a terminal was opened, was not giving me a good feeling. I felt the system was in charge of me, rather than the other way around. 😂😂

2) Deleting everything Conda related meant deleting the recent R. The R I had was from the Ubuntu repository, which is R version 3.6.3 (2020-02-29) - i.e. almost 3 years old.

3) Built Sage with the old R 3.6.3

$ ./configure
$ make -j 50 ptestlong

sage built okay, using the old R. 

When running the doctests I saw this message

"Doctesting 4451 files using 8 threads"

Why does Sage not use 50 threads when running the tests, but only 8? I'd told make to use 50.

The following was the result of the doctesting. The build failed.

sage -t --long --warn-long 41.7 --random-seed=247878169657737533415589408317275108273 src/sage/tests/cmdline.py
    [217 tests, 47.85 s]
----------------------------------------------------------------------
sage -t --long --warn-long 41.7 --random-seed=247878169657737533415589408317275108273 src/sage/rings/polynomial/polynomial_quotient_ring_element.py  # 6 doctests failed
----------------------------------------------------------------------
Total time for all tests: 1898.7 seconds
    cpu time: 12100.3 seconds
    cumulative wall time: 13160.6 seconds
Features detected for doctesting: 4ti2,cvxopt,dvipng,gfan,graphviz,imagemagick,lrslib,nauty,palp,pandoc,pdf2svg,pdftocairo,sage.combinat,sage.geometry.polyhedron,sage.graphs,sage.groups,sage.misc.cython,sage.plot,sage.rings.number_field,sage.rings.padics,sage.rings.real_double,sage.symbolic,sagemath_doc_html,sphinx
pytest is not installed in the venv, skip checking tests that rely on it
make[2]: *** [Makefile:273: test] Error 1
make[2]: Leaving directory '/home/drkirkby/sage-9.8'
make[1]: *** [Makefile:288: ptest] Error 2
make[1]: Leaving directory '/home/drkirkby/sage-9.8'
make: *** [Makefile:294: ptestlong] Error 2
drkirkby@canary:~/sage-9.8$

I expected to see a list of the tests that failed, but I didn't see any list, despite polynomial_quotient_ring_element.py indicated 6 failures.

Next I decided to

1) Build the latest R from source, as I did not want a 3-year old version. That needed quite a bit of work to get all the dependencies R wanted for an optimal configuration. I'm not surprised that maintaining the R package was difficult.

2) Install pytest as there was an error message about it not being installed in the venv, whatever the venv is. (Typo for environment?) That gave me pytest version 4.6.9. I see the latest is 7.x, so I assume 4.6.9 is pretty old

3) Rebuild Sage.

$ ./configure
$ make -j 50 ptestlong.

This time, with pytest present, there's an error message it is not, but the build does not fail. All tests pass. So the rpy in Sage is compatible with the latest R, and does not need updating just to work with the current version of R.


sage -t --long --warn-long 61.6 --random-seed=244047359021345785605807767847544173245 src/sage/libs/flint/fmpz_poly_mat.pxd
    [0 tests, 0.00 s]
sage -t --long --warn-long 61.6 --random-seed=244047359021345785605807767847544173245 src/sage/schemes/toric/sheaf/__init__.py
    [0 tests, 0.00 s]
----------------------------------------------------------------------
All tests passed!
----------------------------------------------------------------------
Total time for all tests: 1873.0 seconds
    cpu time: 12087.9 seconds
    cumulative wall time: 13194.0 seconds
Features detected for doctesting: 4ti2,cvxopt,dvipng,gfan,graphviz,imagemagick,lrslib,nauty,palp,pandoc,pdf2svg,pdftocairo,sage.combinat,sage.geometry.polyhedron,sage.graphs,sage.groups,sage.misc.cython,sage.plot,sage.rings.number_field,sage.rings.padics,sage.rings.real_double,sage.symbolic,sagemath_doc_html,sphinx
pytest is not installed in the venv, skip checking tests that rely on it
make[2]: Leaving directory '/home/drkirkby/sage-9.8

Reply all
Reply to author
Forward
0 new messages