No longer able to build 7.2 or 7.3 betas on Mac OS X El Capitan

[Paul Masson]

In an effort to get past a recurring issue for reviewing Trac #21034, I did "make distclean" followed by "make". Now I am no longer able to build Sage as I have been doing for the last couple months, either the master or develop branch. This is the error message I keep getting:

* package: atlas-3.10.2.p2

  log file: /Users/Masson/Downloads/GitHub/sage/logs/pkgs/atlas-3.10.2.p2.log

  build directory: /Users/Masson/Downloads/GitHub/sage/local/var/tmp/sage/build/atlas-3.10.2.p2

* package: cython-0.24.p0

  log file: /Users/Masson/Downloads/GitHub/sage/logs/pkgs/cython-0.24.p0.log

  build directory: /Users/Masson/Downloads/GitHub/sage/local/var/tmp/sage/build/cython-0.24.p0

The Atlas package is mentioned in Trac #21029.

Any indications of how to get past this, short of deleting the entire repository, would be greatly appreciated. I'm baffled as to why this would stop working.

[Volker Braun]

Whats in the logs?

Did you just apply the most recent OSX update?

[Francois Bissey]

Definitely want the log. By default we don’t build ATLAS on OS X but the spkg is called
to set up the use OS X “vectorize” framework in that case.

François

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

[Paul Masson]

Most recent update applied: running OS X 10.11.6

Install log: https://drive.google.com/open?id=0B9H_P15GCK5yME43Sjk2bmFZSmM

[Paul Masson]

The posted install log was from an attempt to build the master branch.

Volker, I submitted a reply to sage-release a couple days ago stating that 7.3.beta9 built cleanly on my system. That was true before the "make distclean" but clearly no longer applies.

[leif]

Paul Masson wrote:
> Most recent update applied: running OS X 10.11.6

Paul, you really shouldn't post install.log of *parallel* builds (at
least not before Sage 7.3.*; "master" is of course 7.2), but better
those of the packages which failed to build (namely those logs mentioned
in the error message you posted).


> Install log: https://drive.google.com/open?id=0B9H_P15GCK5yME43Sjk2bmFZSmM

TL;DR:

cython-0.24.p0
====================================================
Setting up build directory for cython-0.24.p0
Finished set up
****************************************************
Host system:
Darwin MacBook-Retina.local 15.6.0 Darwin Kernel Version 15.6.0: Thu Jun
23 18:25:34 PDT 2016; root:xnu-3248.60.10~1/RELEASE_X86_64 x86_64
****************************************************
C compiler: gcc
C compiler version:
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/Users/Masson/Downloads/GitHub/sage/local/libexec/gcc/x86_64-apple-darwin15.6.0/4.9.3/lto-wrapper
Target: x86_64-apple-darwin15.6.0
Configured with: ../src/configure
--prefix=/Users/Masson/Downloads/GitHub/sage/local
--with-local-prefix=/Users/Masson/Downloads/GitHub/sage/local
--with-gmp=/Users/Masson/Downloads/GitHub/sage/local
--with-mpfr=/Users/Masson/Downloads/GitHub/sage/local
--with-mpc=/Users/Masson/Downloads/GitHub/sage/local --with-system-zlib
--disable-multilib --disable-nls --enable-languages=c,c++,fortran
--disable-libitm --with-build-config=bootstrap-debug --without-isl
--without-cloog
Thread model: posix
gcc version 4.9.3 (GCC)
...
Checking .pth file support in /Library/Python/2.7/site-packages/
error: can't create or remove files in install directory

The following error occurred while trying to add or remove files in the
installation directory:

[Errno 13] Permission denied:
'/Library/Python/2.7/site-packages/test-easy-install-38731.pth'

The installation directory you specified (via --install-dir, --prefix, or
the distutils default setting) was:

/Library/Python/2.7/site-packages/

Perhaps your account does not have write access to this directory? If the
installation directory is a system-owned directory, you may need to sign in
as the administrator or "root" account. If you do not have administrative
access to this machine, you may wish to choose a different installation
directory, preferably one that is listed in your PYTHONPATH environment
variable.

For information on other options, you may wish to consult the
documentation at:

https://pythonhosted.org/setuptools/easy_install.html

Please make the appropriate changes for your system and try again.

Error installing Cython

real 0m0.415s
user 0m0.180s
sys 0m0.128s
************************************************************************
Error installing package cython-0.24.p0
************************************************************************



The error in building ATLAS is more funny:

atlas-3.10.2.p2
====================================================
Setting up build directory for atlas-3.10.2.p2
Finished set up
Error: The spkg-install script is written in Python, but the Python
package is not yet installed in Sage. You should add $(PYTHON)
as dependency in build/pkgs/atlas/dependencies
make[2]: ***
[/Users/Masson/Downloads/GitHub/sage/local/var/lib/sage/installed/atlas-3.10.2.p2]
Error 1



-leif

> On Sunday, July 24, 2016 at 11:54:06 PM UTC-7, Volker Braun wrote:
>
> Whats in the logs?
>
> Did you just apply the most recent OSX update?
>
>
>
> On Monday, July 25, 2016 at 4:44:21 AM UTC+2, Paul Masson wrote:
>
> In an effort to get past a recurring issue for reviewing Trac
> #21034, I did "make distclean" followed by "make". Now I am no
> longer able to build Sage as I have been doing for the last
> couple months, either the master or develop branch. This is the
> error message I keep getting:
>
> |
>

> *package:atlas-3.10.2.p2

>
>
> log
> file:/Users/Masson/Downloads/GitHub/sage/logs/pkgs/atlas-3.10.2.p2.log
>
>
> build
> directory:/Users/Masson/Downloads/GitHub/sage/local/var/tmp/sage/build/atlas-3.10.2.p2
>
>
>

> *package:cython-0.24.p0

[Paul Masson]

Sorry about that Leif. My bad.

I'm the only person using this machine, so I'm the administrator. The directory in the Cython error is a system directory. 

I think my very first build of 7.2 a couple months ago was not a parallel build: is that an issue for an initial build?

[leif]

Paul Masson wrote:
> Sorry about that Leif. My bad.
>
> I'm the only person using this machine, so I'm the administrator. The
> directory in the Cython error is a system directory.

Yes, and Sage shouldn't try to write to /that/.

> I think my very first build of 7.2 a couple months ago was not a
> parallel build: is that an issue for an initial build?

It shouldn't, but somehow Sage's Python apparently doesn't get built
[early enough]; you may try to build with just one 'make' job ('-j1'),
be it just to better track down what's going wrong.


People on MacOS X will have better suggestions I guess.


-leif

[Paul Masson]

Ran make again without specifying any jobs, which I understand to be the same as specifying one job. Same error from Atlas this time as well. Here's the contents of that log:

Found local metadata for atlas-3.10.2.p2

Attempting to download package atlas-3.10.2.tar.bz2 from mirrors

http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2

[......................................................................]

atlas-3.10.2.p2

====================================================

Setting up build directory for atlas-3.10.2.p2

Finished set up

Error: The spkg-install script is written in Python, but the Python

package is not yet installed in Sage.  You should add $(PYTHON)

as dependency in build/pkgs/atlas/dependencies

 

[leif]

Paul Masson wrote:
> Ran make again without specifying any jobs, which I understand to be the
> same as specifying one job. Same error from Atlas this time as well.
> Here's the contents of that log:
>
> Found local metadata for atlas-3.10.2.p2
> Attempting to download package atlas-3.10.2.tar.bz2 from mirrors
> http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2

> [.......................................................................]
> atlas-3.10.2..p2
> ====================================================
> Setting up build directory for atlas-3.10.2..p2

> Finished set up
> Error: The spkg-install script is written in Python, but the Python
> package is not yet installed in Sage. You should add $(PYTHON)
> as dependency in build/pkgs/atlas/dependencies

Yep. Both errors (in ATLAS and Cython) have the same cause, namely that
Sage's Python doesn't get built. (I know how I could manage to get the
same errors, but in your case something else seems to be wrong.)

Could you post your build/make/Makefile somewhere, too? (It's not that
large, but long, ~2900 lines...)


And to go triple-safe, the output of 'ls local/var/lib/sage/installed/'.

[Travis Scrimshaw]

I don't want to derail the debugging (which has utility), but perhaps try forcing python to build with "make python" (or "sage -i python") just to get a working version?

Best,
Travis

[leif]

leif wrote:
> Paul Masson wrote:
>> Ran make again without specifying any jobs, which I understand to be the
>> same as specifying one job. Same error from Atlas this time as well.
>> Here's the contents of that log:
>>
>> Found local metadata for atlas-3.10.2.p2
>> Attempting to download package atlas-3.10.2.tar.bz2 from mirrors
>> http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2
>> [.......................................................................]
>> atlas-3.10.2..p2
>> ====================================================
>> Setting up build directory for atlas-3.10.2..p2
>> Finished set up
>> Error: The spkg-install script is written in Python, but the Python
>> package is not yet installed in Sage. You should add $(PYTHON)
>> as dependency in build/pkgs/atlas/dependencies
>
> Yep. Both errors (in ATLAS and Cython) have the same cause, namely that
> Sage's Python doesn't get built. (I know how I could manage to get the
> same errors, but in your case something else seems to be wrong.)

P.S.: If you just want to get past this error, try

make python2

and if that succeeded (Python got built), run 'make' (or 'make -j8') again.


-leif

[Paul Masson]

MacBook-Retina:~ Masson$ cd Downloads/GitHub/sage/local/var/lib/sage/installed

MacBook-Retina:installed Masson$ ls

arb-2.8.1.p0 gcc-4.9.3.p1 mpfr-3.1.4 patch-2.7.5

bzip2-1.0.6-20150304 gf2x-1.1.p1 mpir-2.7.2 pkgconf-0.9.7

flint-2.5.2.p0 mpc-1.0.3.p0 ntl-9.6.2.p1

[John H Palmieri]

What do the commands

which python
python --version

tell you? OS X should include a version of Python. Mine has version 2.7.10, dating from last December, in /usr/bin. If yours is missing, maybe your PATH is wrong or maybe your Xcode installation is broken somehow.

[Paul Masson]

MacBook-Retina:~ Masson$ which python

/usr/bin/python

MacBook-Retina:~ Masson$ python --version

Python 2.7.10

[Jeroen Demeyer]

It seems like something is wrong with make. What does "make --version" give?

Do you have any unusual environment variables set? Can you show the
output of "set"?

[Jeroen Demeyer]

Got it! Your Makefile is corrupted, see this part:

[...]
inst_pkgconf = $(INST)/pkgconf-0.9.7
inst_pkgconfig = $(INST)/pkgconfig-1.1.0
inst_planarity = $(INST)/planarity-2.2.0
inst_plantri = $(INST)/plantri-4.5
inst_polytopes_db = $(INST)/polytopes_db-20120220
inst_ppl = $(INST)/ppl-1.1


# All standard packages
[...]

This list unexpectedly ends at ppl, which means that any package which
comes alphabetically after ppl will not be installed.

I will need to see the top-level file config.log to debug this further.

[leif]

Paul Masson wrote:
>
> MacBook-Retina:~ Masson$ cd
> Downloads/GitHub/sage/local/var/lib/sage/installed
>
> MacBook-Retina:installed Masson$ ls
>
> arb-2.8.1.p0gcc-4.9.3.p1mpfr-3.1.4patch-2.7.5
>
> bzip2-1.0.6-20150304gf2x-1.1.p1mpir-2.7.2pkgconf-0.9.7
>
> flint-2.5.2.p0mpc-1.0.3.p0ntl-9.6.2.p1

So that's ok.

But the generated parts of your build/make/Makefile lack a lot of
packages (starting with "pt" to "z" *), so the bug is already in
`configure`.

I'd suggest to save a copy of your current build/make/Makefile somewhere
for later comparison, rerun ./configure, and see whether you get a
different (especially longer) build/make/Makefile after doing so.

(If it hasn't changed, there's no point in rerunning 'make'.)


-leif

__________
* every package after PPL, i.e. ptyprocess pybtex pycrypto pyflakes
pygments pynac pyopenssl pyparsing python2 python3 python_igraph
python_openid pytz pyzmq qepcad r r_jupyter ratpoints readline rpy2
rubiks rw saclib sage_mode sagenb sagetex scipy scons setuptools
setuptools_scm simplegeneric singledispatch singular six snowballstemmer
speaklater sphinx sqlalchemy sqlite symmetrica sympow sympy tachyon
tdlib termcap threejs tides topcom tornado trac traitlets twisted
valgrind vcversioner werkzeug zeromq zlib zn_poly zope_interface

[leif]

Jeroen Demeyer wrote:
> Got it! Your Makefile is corrupted,

I wouldn't say it's "corrupted", it's misgenerated: All packages
(alphabetically) starting with ptyprocess are /consistently/ missing.

So apparently already the loop in filtered_packages_list()

for DIR in $SAGE_ROOT/build/pkgs/*;

consistently ends too early, as filtered_packages_list() is later called
a couple of times.


-leif

[leif]

leif wrote:
> Jeroen Demeyer wrote:
>> Got it! Your Makefile is corrupted,
>
> I wouldn't say it's "corrupted", it's misgenerated: All packages
> (alphabetically) starting with ptyprocess are /consistently/ missing.
>
> So apparently already the loop in filtered_packages_list()
>
> for DIR in $SAGE_ROOT/build/pkgs/*;
>
> consistently ends too early, as filtered_packages_list() is later called
> a couple of times.

P.S.:

The problematic thing here is that 'make' doesn't bail out with "Don't
know how to make target 'foo'. Stop." as one would expect, but due to
the way our Makefile is written. dependencies simply get dropped, as
undefined variables evaluate to the empty string.


P.P.S.: I think I know where the bug in 'configure[.ac]' is (it's
actually multiple instances of the same)...

[Paul Masson]

MacBook-Retina:~ Masson$ make --version

GNU Make 3.81

Copyright (C) 2006  Free Software Foundation, Inc.

This is free software; see the source for copying conditions.

There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A

PARTICULAR PURPOSE.

This program built for i386-apple-darwin11.3.0

----------------

I have never consciously changed any environment variables. Here's the current configuration:

MacBook-Retina:sage Masson$ sage --sh -c env

AS=as

LDFLAGS=-L/Users/Masson/Downloads/GitHub/sage/local/lib -Wl,-rpath,/Users/Masson/Downloads/GitHub/sage/local/lib 

AR=ar

SAGE_LOGS=/Users/Masson/Downloads/GitHub/sage/logs/pkgs

TERM_PROGRAM=Apple_Terminal

UNAME=Darwin

ECLDIR=/Users/Masson/Downloads/GitHub/sage/local/lib/ecl/

TERM=xterm-256color

SHELL=/bin/bash

OBJC=/usr/bin/cc

OBJCXX=/usr/bin/c++

TMPDIR=/var/folders/np/_r579ck92g1f9x1tzcvdhy5m0000gn/T/

SAGE_ENV_SOURCED=4

Apple_PubSub_Socket_Render=/private/tmp/com.apple.launchd.NtemTgLxXJ/Render

LIBRARY_PATH=/Users/Masson/Downloads/GitHub/sage/local/lib

MKDIR=mkdir

SAGE_DOC_MATHJAX=True

TERM_PROGRAM_VERSION=361.1

SAGE_ORIG_PATH=/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/Masson/Downloads/GitHub/git-trac-command/bin:/Users/Masson/Downloads/GitHub/sage:/opt/X11/bin

PYTHON_EGG_CACHE=/Users/Masson/.sage//.python-eggs

CYSIGNALS_CRASH_LOGS=/Users/Masson/.sage//crash_logs

TERM_SESSION_ID=5D0FED3F-3132-4569-A9AB-513DAE39155E

SAGE_REPO_AUTHENTICATED=ssh://g...@trac.sagemath.org:2222/sage.git

SAGE_NUM_THREADS=1

DOT_SAGE=/Users/Masson/.sage/

PYTHONUSERBASE=/Users/Masson/.sage//local

USER=Masson

SAGE_SCRIPTS_DIR=/Users/Masson/Downloads/GitHub/sage/src/bin

SINGULAR_EXECUTABLE=/Users/Masson/Downloads/GitHub/sage/local/bin/Singular

SSH_AUTH_SOCK=/private/tmp/com.apple.launchd.WO56zkhqPv/Listeners

CPATH=/Users/Masson/Downloads/GitHub/sage/local/include

__CF_USER_TEXT_ENCODING=0x1F5:0x0:0x0

SAGE_LOCAL=/Users/Masson/Downloads/GitHub/sage/local

MAXIMA_PREFIX=/Users/Masson/Downloads/GitHub/sage/local

SAGE_REPO_ANONYMOUS=git://trac.sagemath.org/sage.git

SAGE_EXTCODE=/Users/Masson/Downloads/GitHub/sage/local/share/sage/ext

SAGE_SPKG_INST=/Users/Masson/Downloads/GitHub/sage/local/var/lib/sage/installed

SAGE_STARTUP_FILE=/Users/Masson/.sage//init.sage

CPP=cpp

SAGE_DOC_SRC=/Users/Masson/Downloads/GitHub/sage/src/doc

SAGE_DISTFILES=/Users/Masson/Downloads/GitHub/sage/upstream

PATH=/Users/Masson/Downloads/GitHub/sage/build/bin:/Users/Masson/Downloads/GitHub/sage/src/bin:/Users/Masson/Downloads/GitHub/sage/local/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/Masson/Downloads/GitHub/git-trac-command/bin:/Users/Masson/Downloads/GitHub/sage:/opt/X11/bin

R_MAKEVARS_USER=/Users/Masson/.sage//R/Makevars.user

SAGE_ETC=/Users/Masson/Downloads/GitHub/sage/local/etc

LD=ld

PWD=/Users/Masson/Downloads/GitHub/sage

F90=gfortran

MV=mv

LANG=en_US.UTF-8

__sage__=

F95=gfortran

SAGE_ROOT=/Users/Masson/Downloads/GitHub/sage

XPC_FLAGS=0x0

SAGE_SHARE=/Users/Masson/Downloads/GitHub/sage/local/share

F77=gfortran

CP=cp

CCACHE_BASEDIR=/Users/Masson/Downloads/GitHub/sage

LN=ln

XPC_SERVICE_NAME=0

SAGE_ORIG_PATH_SET=True

CXX=g++

HOME=/Users/Masson

SHLVL=2

MAKE=make

TERMINFO=/Users/Masson/Downloads/GitHub/sage/local/share/terminfo

FC=gfortran

LOGNAME=Masson

SAGE_SRC=/Users/Masson/Downloads/GitHub/sage/src

MACOSX_DEPLOYMENT_TARGET=10.9

CYSIGNALS_CRASH_DAYS=7

GPHELP=/Users/Masson/Downloads/GitHub/sage/local/bin/gphelp

SAGE_DOC=/Users/Masson/Downloads/GitHub/sage/local/share/doc/sage

RANLIB=ranlib

DISPLAY=/private/tmp/com.apple.launchd.iqsFS6XFtS/org.macosforge.xquartz:0

SINGULARPATH=/Users/Masson/Downloads/GitHub/sage/local/share/singular

CC=gcc

CHMOD=chmod

SAGE_NUM_THREADS_PARALLEL=4

MACOSX_VERSION=15

GP_DATA_DIR=/Users/Masson/Downloads/GitHub/sage/local/share/pari

TOUCH=touch

GPDOCDIR=/Users/Masson/Downloads/GitHub/sage/local/share/pari/doc

SAGE64=no

_=/usr/bin/env

[Paul Masson]

[Paul Masson]

Running "configure" resulted in a makefile that terminates at the same package. Here's some possibly relevant error messages from the configure process:

    planarity-2.2.0

    plantri-4.5

    polytopes_db-20120220

"/Users/Masson/Downloads/GitHub/sage/build/pkgs/prompt_toolkit/type" is missing.

    ppl-1.1

"/Users/Masson/Downloads/GitHub/sage/build/pkgs/prompt_toolkit/type" is missing.

"/Users/Masson/Downloads/GitHub/sage/build/pkgs/prompt_toolkit/type" is missing.

"/Users/Masson/Downloads/GitHub/sage/build/pkgs/prompt_toolkit/type" is missing.

"/Users/Masson/Downloads/GitHub/sage/build/pkgs/prompt_toolkit/type" is missing.

checking that generated files are newer than configure... done

configure: creating ./config.status

[Paul Masson]

The error message here led me to look at the "prompt_toolkit" package. That directory is empty, as is one for "wcwidth". Removing both folders resulted in a makefile with all actual packages. Running make again now.

The supposed packages "prompt_toolkit" and "wcwidth" aren't part of either the current master or develop branch. Could they have wandered in from some old Trac ticket I reviewed?

[Volker Braun]

You tried the ipython 5.0 ticket, I presume.

This left an empty directory lying around. Git doesn't know about empty directories, but our build system assumes that it is a package. And then fails and writes an incomplete makefile. You need to delete build/pkgs/prompt_toolkit, either by hand or via "git clean -fdx" (the latter will throw away everything that isn't checked in).

[Paul Masson]

As I recall I did checkout that ticket (21006) by accident when I wanted one (21008) a couple digits away. That's the source of the problem.

Why can't the build system recognize an empty folder and pass over it?

[leif]

Paul Masson wrote:
> As I recall I did checkout that ticket (21006) by accident when I wanted
> one (21008) a couple digits away. That's the source of the problem.
>
> Why can't the build system recognize an empty folder and pass over it?

That's a genuine bug in Sage's 'configure'. Not only that it doesn't
ignore empty folders, but doesn't even bail out with an error as
intended, since the 'exit 1' terminates the loop / function, but not the
'configure' script (hence just the package list gets truncated, and
'configure' later exits normally).

I'll open a ticket unless someone beats me in doing so. (IMHO worth a
blocker, but I don't know how far Volker is in preparing the 7.3 release
candidate.)


-leif

> > > Attempting to download package atlas-3.10.2.tar..bz2

> from mirrors
> > >
> >
> http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2
> <http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2>
>
> >

> <http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2
> <http://mirrors.xmission.com/sage/spkg/upstream/atlas/atlas-3.10.2.tar.bz2>>
>
> >
> > >
> >
> [........................................................................]

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

> <mailto:sage-devel+...@googlegroups.com>.

> To post to this group, send email to sage-...@googlegroups.com

> <mailto:sage-...@googlegroups.com>.

[Paul Masson]

Success! Many thanks to all who contributed.

[leif]

Paul Masson wrote:
> Success! Many thanks to all who contributed.

Fine.

Just for the record: While 'git status' doesn't show untracked empty
folders as Volker mentioned,

git ls-files -o --directory --exclude-standard

would have shown you (also) the left-over folders which led to this.

("List other (untracked) files which aren't (explicitly) ignored." The
'--directory' is necessary to list also *empty* folders of that kind,
while its actual meaning is to /not show the contents/ of matching
directories, but only themselves, similar to 'ls -d ...'.)


But as I said, there's definitely a bug in Sage's top-level 'configure'
we will fix, such that this kind of obscure error will never happen again.


-leif

[leif]

leif wrote:
> Paul Masson wrote:
>> As I recall I did checkout that ticket (21006) by accident when I wanted
>> one (21008) a couple digits away. That's the source of the problem.
>>
>> Why can't the build system recognize an empty folder and pass over it?
>
> That's a genuine bug in Sage's 'configure'. Not only that it doesn't
> ignore empty folders, but doesn't even bail out with an error as
> intended, since the 'exit 1' terminates the loop / function, but not the
> 'configure' script (hence just the package list gets truncated, and
> 'configure' later exits normally).
>
> I'll open a ticket unless someone beats me in doing so. (IMHO worth a
> blocker, but I don't know how far Volker is in preparing the 7.3 release
> candidate.)

Done. Needs review:

https://trac.sagemath.org/ticket/21102