The problem went away after I did "make distclean; make build". Originally what I did after checking out a new branch was (1) put the new tarball into upstream/ (2) edit the file build/pkgs/eclib/package-version.txt (3) issue the magic command which updates the checksum (4) ./sage -i eclib [causes the new eclib to be built OK] (5) make build [rebuilds some cython files which depend on eclib] (6) run tests
Surely that should work, so whatever is deciding what needs doing in step (5) is possibly not getting it quite right. John On Wed, 27 Feb 2019 at 07:35, Frédéric Chapoton <fchapot...@gmail.com> wrote: > This problem is also seen in the python3 patchbot report, as can be found > here : > > > https://patchbot.sagemath.org/log/0/Ubuntu/16.04/x86_64/4.4.0-141-generic/atlas/2019-02-24%2011:31:45?short > > Frederic > > Le mardi 26 février 2019 17:52:39 UTC+1, John Cremona a écrit : >> >> I am testing a new version of the standard spkg eclib, see >> https://trac.sagemath.org/ticket/27360. I am getting doctesting >> failures caused *only* because of changes in output buffering resulting in >> the tail end of the output of one command appearing at the beginning of the >> output of the next. For example if you do >> >> sage: E = mwrank_EllipticCurve([0,0,1,-7,6]) >> sage: EQ = mwrank_MordellWeil(E) >> sage: EQ.search(2) >> >> as in src/sage/libs/eclib/interface.py near the end, then the last of >> those commands causes eclib (which is a C++ library) to output some >> progress to stdout. In that files there are several examples of that sort, >> causing output to stdout as a side-effect. In 8.7.beta5 (and all previous >> versions including many previous versions of eclib) all is well, but after >> building the new eclib version (which makes no relevant changes whatsoever) >> I find that half of the last line of the side-effect output is not being >> sent to the output stream until the next doctest is run, with the result >> that both test fail since the output is not what was expected. >> >> I based my working branch on 8.7.beta5 which works fine. I tried adding >> some sys.stdout.flush() commands in the code and/or the doctests but it >> makes no difference -- presumably eclib's stdout is not the same channel as >> python's. >> >> Any suggestions other than "you must have made some change in eclib which >> caused this"? Since I didn't!? >> >> John >> >> >> -- > 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+unsubscr...@googlegroups.com. > To post to this group, send email to sage-devel@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-devel. > For more options, visit https://groups.google.com/d/optout. > -- 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+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
