On Sun, Apr 29, 2018 at 11:55 AM, Erik Bray <[email protected]> wrote: > On Sun, Apr 29, 2018 at 11:47 AM, Erik Bray <[email protected]> wrote: >> On Sun, Apr 29, 2018 at 11:02 AM, Erik Bray <[email protected]> wrote: >>> On Sun, Apr 29, 2018 at 10:53 AM, Erik Bray <[email protected]> wrote: >>>> On Sun, Apr 29, 2018 at 4:40 AM, Nathan Dunfield <[email protected]> >>>> wrote: >>>>> On Saturday, April 28, 2018 at 7:41:49 PM UTC-5, Erik Bray wrote: >>>>>> >>>>>> I just finished a full build from scratch on Ubuntu 18.04 and it went >>>>>> fine. This was of 8.2.rc4. >>>>>> >>>>>> I don't recall off the top of my head whether anything changed between >>>>>> rc1 and rc4 that might have effected this, but you might try bumping a >>>>>> rc4 just in case. >>>>> >>>>> >>>>> Erik, >>>>> >>>>> First, thanks for looking into this. I erred when I wrote rc1, I too was >>>>> working with the very latest rc4. I was using my own variant of your >>>>> original Dockerfiles, but I just did another run where I cloned >>>>> "sagemath/docker-images", edited the first line of >>>>> "sagemath-base/Dockerfile" and then did: >>>>> >>>>> docker build --tag="sagemath/sagemath-base" sagemath-base >>>>> docker build --tag="sagemath/sagemath-develop" sagemath-develop >>>>> >>>>> >>>>> The error log looks pretty much the same as before except this time the >>>>> package was "ecm-7.0.4.p0" rather than "ecm-7.0.4.p1". >>>> >>>> Interesting--when I did the build I did it by hand, though for lack of >>>> better build instructions I basically just manually followed the steps >>>> in the Dockerfile. I will try again the way you did it. >>> >>> Are you sure you were building 8.2.rc4? In the Dockerfile the default >>> is (perhaps wrongly, in retrospect) to build from Sage's "master" >>> branch, rather than the more current "develop" branch (this is true >>> even for the "sagemath-develop" Dockerfile, which is confusing). >>> >>> The version of the ecm package in 8.2.rc4 should be 7.0.4.p1. >> >> Ah, well regardless I just got the failure in ECM too while running >> >> docker build --tag="sagemath/sagemath-develop" --build-arg >> SAGE_BRANCH=develop sagemath-develop >> >> If I had to wager a guess, the only difference between the Dockerfile >> and when I tried building manually is the use of SAGE_FAT_BINARY=yes. > > Yep, SAGE_FAT_BINARY=yes is what did it. I'll look into it further.
This appears to be related, and explains why we're just seeing this on bionic (actually introduced in 16.10, but the Docker images are currently based on 16.04). -- 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 [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
