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? On Tuesday, July 26, 2016 at 1:20:38 PM UTC-7, Volker Braun wrote: > > 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). > > > > On Tuesday, July 26, 2016 at 9:55:48 PM UTC+2, Paul Masson wrote: >> >> 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 >> >> >> On Tuesday, July 26, 2016 at 6:31:36 AM UTC-7, leif wrote: >>> >>> 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 >>> >>> >>> > On Monday, July 25, 2016 at 8:26:25 PM UTC-7, 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 >>> > < >>> 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/'. >>> > >>> > >>> > -leif >>> > >>> > > On Monday, July 25, 2016 at 4:47:06 PM UTC-7, leif wrote: >>> > > >>> > > 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 >>> >>> >>> -- 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.
