Sage 9.1.rc4 released!

[Volker Braun]

As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html 

Now would be a good time for you to test ;-)

c5c78afdfd (tag: 9.1.rc4, trac/develop) Updated SageMath version to 9.1.rc4

f2d714d003 Trac #25206: Update README.md and Installation Manual for 9.1, integrate instructions for Windows (WSL and Cygwin) in manuals

632cfcebc3 Trac #29443: Update scipy to 1.2.3 (last release supporting py2)

53b3703897 Trac #29622: cygwin, conda, homebrew: Fixes for CI

bfbaddcc58 (tag: 9.1.rc3) Updated SageMath version to 9.1.rc3

[Samuel Lelièvre]

> From: Volker Braun
> To: sage-release
> Date: Sat 2020-05-09 10:22 UTC
> Subject: [sage-release] Sage 9.1.rc4 released!

>
>
> Now would be a good time for you to test ;-)

Thanks for this new release. Tested on two systems.

## Debian 9

```
$ make ptestlong
[...]
----------------------------------------------------------------------
sage -t --long src/sage/matrix/matrix2.pyx # Timed out
----------------------------------------------------------------------
Total time for all tests: 2194.0 seconds
cpu time: 13681.4 seconds
cumulative wall time: 14915.5 seconds
Makefile:182: recipe for target 'ptestlong' failed
make: *** [ptestlong] Error 4
$
```

The test passed when run standalone.

```
$ ./sage -t --long src/sage/matrix/matrix2.pyx
too many failed tests, not using stored timings
Running doctests with ID 2020-05-10-01-39-00-fc8987f7.
Git branch: develop
Using --optional=build,dochtml,memlimit,sage
Doctesting 1 file.
sage -t --long src/sage/matrix/matrix2.pyx
[2574 tests, 14.79 s]
----------------------------------------------------------------------
All tests passed!
----------------------------------------------------------------------
Total time for all tests: 15.4 seconds
cpu time: 16.1 seconds
cumulative wall time: 14.8 seconds
```

## macOS 10.14.6 Mojave

With a fair set of packages installed via Homebrew,
but not a comprehensive one. I might try again with
more brew-installed packages.

```
$ make ptestlong
[...]
----------------------------------------------------------------------
All tests passed!
----------------------------------------------------------------------
Total time for all tests: 14776.3 seconds
cpu time: 23110.2 seconds
cumulative wall time: 42792.6 seconds
```

During `make ptestlong`,
- two "Python Crash" dialogue windows popped up,
offering to report the crash to Apple.
I saved the crash logs in case it's any use.
- one "Java Update" dialogue window popped up,
offering to install the latest Java update.

[Enrique Artal]

I am using Fedora 32 with gcc 10. Upgrading to rc3 worked because apparently  no incompatible package changed. I could not upgrade to rc4 and because of scipy, trying to downgrade to rc3 didn't work either. After a dist-clean I tried with the gcc package to avoid those problems but it did not compile. Is there a workaround? Best, Enrique.

[Dima Pasechnik]

you can revert the commit upgrading scipy, I suppose this will do the trick.

--
You received this message because you are subscribed to the Google Groups "sage-release" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-release...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/b4405a9c-a33a-4949-b6c8-a161d228301f%40googlegroups.com.

[Matthias Köppe]

On Sunday, May 10, 2020 at 11:29:08 AM UTC-7, Enrique Artal wrote:

I am using Fedora 32 with gcc 10. Upgrading to rc3 worked because apparently  no incompatible package changed. I could not upgrade to rc4 and because of scipy, trying to downgrade to rc3 didn't work either. After a dist-clean I tried with the gcc package to avoid those problems but it did not compile. Is there a workaround? Best, Enrique.

Fedora 32 / GCC 10 is not supported; see https://trac.sagemath.org/ticket/29456. 

You could try compiling with "./configure --without-system-gcc". Then Sage will build its own copy of gcc 9.

[Dima Pasechnik]

On Sun, May 10, 2020 at 7:51 PM Matthias Köppe <matthia...@gmail.com> wrote:

>
> On Sunday, May 10, 2020 at 11:41:17 AM UTC-7, Dima Pasechnik wrote:
>>
>> you can revert the commit upgrading scipy, I suppose this will do the trick.
>>>
>>>
>

> No, this has nothing to do with scipy.
>
I read the message in question as if it was workfing with rc3. Anyway.

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

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/d408344f-efee-4dcd-b2b9-3aa4087693e6%40googlegroups.com.

[Enrique Artal]

I tried, but the compilation of gcc failed. I will try again to put the logs.

[Matthias Köppe]

On Sunday, May 10, 2020 at 12:00:09 PM UTC-7, Dima Pasechnik wrote:

On Sun, May 10, 2020 at 7:51 PM Matthias Köppe <matthia...@gmail.com> wrote:
>
> On Sunday, May 10, 2020 at 11:41:17 AM UTC-7, Dima Pasechnik wrote:
>>
>> you can revert the commit upgrading scipy, I suppose this will do the trick.
>>>
>>>
>
> No, this has nothing to do with scipy.
>
I read the message in question as if it was workfing with rc3. Anyway.

Sorry, actually I have to correct myself. All versions of scipy that we can use for 9.1 do not work with GCC 10. We need to upgrade to new scipy that is Python 3 only. See https://trac.sagemath.org/ticket/29425

 

[Enrique Artal]

Two of the failed packages, palp and sympow exist in Fedora 32. Does it help?

El domingo, 10 de mayo de 2020, 20:50:33 (UTC+2), Matthias Köppe escribió:

[Matthias Köppe]

On Sunday, May 10, 2020 at 12:06:18 PM UTC-7, Enrique Artal wrote:

Two of the failed packages, palp and sympow exist in Fedora 32. Does it help?

We do not have spkg-configure.m4 for these packages yet. These would need to be added. See https://trac.sagemath.org/ticket/27330

 

[Enrique Artal]

Last version of scipy was most probably compiled before upgrading to Fedora 32

El domingo, 10 de mayo de 2020, 21:00:09 (UTC+2), Dima Pasechnik escribió:

On Sun, May 10, 2020 at 7:51 PM Matthias Köppe <matthia...@gmail.com> wrote:
>
> On Sunday, May 10, 2020 at 11:41:17 AM UTC-7, Dima Pasechnik wrote:
>>
>> you can revert the commit upgrading scipy, I suppose this will do the trick.
>>>
>>>
>
> No, this has nothing to do with scipy.
>
I read the message in question as if it was workfing with rc3. Anyway.
> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to sage-r...@googlegroups.com.

[Dima Pasechnik]

I just opened tickets for these, should be ready by tomorrow... :-)

>
> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to sage-release...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/a6ce0c89-9234-4393-8236-003de2808cab%40googlegroups.com.

[Matthias Köppe]

On Sunday, May 10, 2020 at 12:35:27 PM UTC-7, Enrique Artal wrote:

Last version of scipy was most probably compiled before upgrading to Fedora 32

Yes, that would be my guess too. 

[Matthias Köppe]

On Sunday, May 10, 2020 at 12:03:41 PM UTC-7, Enrique Artal wrote:

I tried, but the compilation of gcc failed. I will try again to put the logs.

 

Ah ok, now I see that you had said that in your original message.

I think the best course of action to support Fedora 32 in Sage 9.1 would be to try to fix the gcc build. 

But I'm not sure if we want to hold up the 9.1 release for that.

An alternative would be to do a very short release cycle for Sage 9.2 which removes Python 2 support and does the necessary package updates to support GCC 10. If Fedora has packages for palp and sympow, it must have patches that we can apply as well.

[Enrique Artal]

I am not in a hurry, I have working copies of sage in other computers with network access, and in this laptop I can use the develop branch. I can test any solution. Best, E.

[Matthias Köppe]

On Sunday, May 10, 2020 at 12:44:01 PM UTC-7, Matthias Köppe wrote:

I think the best course of action to support Fedora 32 in Sage 9.1 would be to try to fix the gcc build. 

But I'm not sure if we want to hold up the 9.1 release for that.

I have created https://trac.sagemath.org/ticket/29674 (fedora-32: Building gcc (9.2.0) spkg fails with gcc 10) for this

[Sébastien Labbé]

Testing many optional and external packages, I get

Using --optional=4ti2,cbc,ccache,cryptominisat,dot2tex,e_antic,external,fricas,glucose,latte_int,lidia,lrslib,

memlimit,normaliz,notedown,openssl,pandoc_attributes,pycosat,pynormaliz,rst2ipynb,sage,sage_numerical_backends_coin,

sage_numerical_backends_cplex,sage_numerical_backends_gurobi

----------------------------------------------------------------------
sage -t --long src/sage/databases/findstat.py  # 8 doctests failed
sage -t --long src/sage/geometry/polyhedron/base.py  # Bad exit: 1
sage -t --long src/sage/graphs/generators/smallgraphs.py  # 2 doctests failed
----------------------------------------------------------------------
External software detected for doctesting: cplex,ffmpeg,graphviz,imagemagick,internet,latex,pandoc

Rerunning failed test, I get the same. Follow up at https://trac.sagemath.org/ticket/25536

Other than the bad exit in the polyhedron/base.py file, I also get a deprecation warning:

sage: _ = P.affine_hull() ## line 9278 ##

...

DeprecationWarning: affine_hull is deprecated. Please use affine_hull_projection instead.
See https://trac.sagemath.org/29326 for details.

[Jonathan Kliem]

The deprecation warning is the tested output in polyhedron/base.py line 9278.

affine_hull was deprecated in https://trac.sagemath.org/ticket/29326.

I don't see why this test shouldn't pass as it uses the same syntax as other deprecation tests (e.g. in src/sage/matrix/matrix.pyx line 8).

If it really doesn't pass, it should be fixed.

[Sébastien Labbé]

On Monday, May 11, 2020 at 12:10:41 PM UTC+2, Jonathan Kliem wrote:

The deprecation warning is the tested output in polyhedron/base.py line 9278.

affine_hull was deprecated in https://trac.sagemath.org/ticket/29326.

I don't see why this test shouldn't pass as it uses the same syntax as other deprecation tests (e.g. in src/sage/matrix/matrix.pyx line 8).

If it really doesn't pass, it should be fixed.

That machine on which I run the tests does not have enough memory to test polyhedron/base.py properly, so it prints that long log of tests run before explosion and bad exit. Thinking about it again, in fact, I think it is fine, the full output of all tests are shown because of the bad exit. So it is not an error, it must be just the normal output.

Sébastien

[Justin C. Walker]

> On May 9, 2020, at 03:22 , Volker Braun <vbrau...@gmail.com> wrote:
>
> As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html
>
> Now would be a good time for you to test ;-)

Built from fresh clones of the develop branch, without problems, on three macOS systems: 10.13.6 (2017 iMac Pro, 18-core Xeon W); 10.14.6 (2017 MBP, Quad-core Core i7); and 10.15.4 (2019 MBP, 9-core Core i9).

All tests (‘ptestlong’) passed on all systems.

Justin

--
Justin C. Walker
Director
Institute for the Enhancement of the Director's Income
--
Fame is fleeting, but obscurity
just drags on and on. F&E

[Enrique Artal]

Thanks!

El domingo, 10 de mayo de 2020, 21:35:45 (UTC+2), Dima Pasechnik escribió:

On Sun, May 10, 2020 at 8:10 PM Matthias Köppe <matthia...@gmail.com> wrote:
>
> On Sunday, May 10, 2020 at 12:06:18 PM UTC-7, Enrique Artal wrote:
>>
>> Two of the failed packages, palp and sympow exist in Fedora 32. Does it help?
>>>
>>>
>
> We do not have spkg-configure.m4 for these packages yet. These would need to be added. See https://trac.sagemath.org/ticket/27330
>
I just opened tickets for these, should be ready by tomorrow... :-)

>
> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to sage-r...@googlegroups.com.

[Matthias Köppe]

I have completed testing of the fix on this ticket. Needs review so it can hopefully be merged in 9.1. 

[Enrique Artal]

I am trying to check this thicket and also #29672. This is the output of git trac checkout 29674. What am I doing wrong?

 
Loading ticket #29674...
Checking out Trac #29674 remote branch u/mkoeppe/fedora_32__building_gcc__9_2_0__spkg_fails_with_gcc_10 -> local branch t/29674/fedora_32__building_gcc__9_2_0__spkg_fails_with_gcc_10...
Traceback (most recent call last):
 File "/home/artal/.local/bin/git-trac", line 17, in <module>
   cmdline.launch()
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/cmdline.py", line 239, in launch
   app.checkout(args.ticket_or_branch, args.branch_name)
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/app.py", line 118, in checkout
   self._checkout_ticket(int(ticket_or_branch), branch_name)
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/app.py", line 146, in _checkout_ticket
   self.repo.checkout_new_branch(ticket.branch, branch)
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/git_repository.py", line 135, in checkout_new_branch
   self.git.fetch('trac', remote)
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/git_interface.py", line 340, in meth
   return self.execute(git_cmd, *args, **kwds)
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/git_interface.py", line 325, in execute
   result = self._run(cmd, args, kwds,
 File "/home/artal/.local/lib/python3.8/site-packages/git_trac/git_interface.py", line 262, in _run
   raise GitError(result)
git_trac.git_error.GitError: git returned with non-zero exit code (128) when executing "git fetch trac u/mkoeppe/fedora_32__building_gcc__9_2_0__spkg_fails_with_gcc_10"
   STDERR: fatal: 'trac' does not appear to be a git repository
   STDERR: fatal: Could not read from remote repository.
   STDERR:  
   STDERR: Please make sure you have the correct access rights
   STDERR: and the repository exists.

[Matthias Köppe]

On Tuesday, May 12, 2020 at 9:51:15 AM UTC-7, Enrique Artal wrote:

I am trying to check this thicket and also #29672. This is the output of git trac checkout 29674. What am I doing wrong?

Check that the "trac" remote is configured correctly - 

https://doc.sagemath.org/html/en/developer/git_trac.html#git-and-trac-configuration 

[Enrique Artal]

Now I read it more carefully, I did the config in the wrong directory

[Enrique Artal]

It worked for me too!

El martes, 12 de mayo de 2020, 7:55:04 (UTC+2), Matthias Köppe escribió: