Hi Gokhan, On Thu, Jan 28, 2010 at 1:24 PM, Gokhan Sever <gokhanse...@gmail.com> wrote:
<SNIP> > Could someone register me at the Trac? I am getting a bizarre error > there: Please send an email to William Stein, reporting the issue and/or request an account. > Also what is the difference between running the test suite (from > $SAGE_ROOT/README.TXT) and testing the installation with ./sage - > testall > > If you want to run the test suite for each individual spkg as it is > installed, type > > export SAGE_CHECK="yes" This environment variable is for running the test suite as contained in the upstream source code. The idea is that once an spkg is successfully compiled, if you have exported the above environment variable, then the installation process would continue with running the test suite in the upstream project. The source code of the upstream project is contained in the directory src/ of an spkg, so the test suite of that project should also be under src/. The script spkg-check in the top-level directory of the spkg is for running the test suite in the upstream project. See this section of the Developers' Guide for more information on the script spkg-check: http://www.sagemath.org/doc/developer/producing_spkgs.html > Once there is an error built is stopped with this. As I described in > this message: > http://groups.google.com/group/sage-devel/browse_thread/thread/4bef86f4eabac321# > To me, this shouldn't stop the build process instead should be > reported at the end of the action. If you do this export SAGE_CHECK="yes" before running "make" to start the build process, the test suite of any successfully compiled spkg would be run. If an error occurred during the running of any test suite, then the whole build process would be stopped. If you don't want to run the test suite (of the upstream project), don't export this environment variable. However, I do understand that the README.txt doesn't sufficiently document the gravity of the situation you would get yourself into when you do export SAGE_CHECK="yes" before issuing "make". Feel free to open a ticket to better document this in README.txt. > One more, can we test both test_suite and doctests seperated for each > package instead of one-run run-all approach? That is possible, provided that you have successfully compiled Sage. A general process would be to first compile Sage using the command make, and don't do export SAGE_CHECK="yes" prior to issuing make. Once Sage is successfully compiled, you could run the doctests in the Sage library and standard documentation as documented in the section "Parallel Testing the Sage Library" of the Developers' Guide: http://www.sagemath.org/doc/developer/doctesting.html After running the doctests, you could then run the test suite in each of the spkg's. To do so, you could do export SAGE_CHECK="yes" and then force a re-installation of an spkg using the command ./sage -f /path/or/URL/to/spkg This would re-install the spkg. Once the spkg is successfully re-compiled, its test suite would be run. A downside to this approach is that it's manual and tedious for running the test suites of all spkg's: you need to force a re-installation for each spkg. But you could write a Bash script, say, to automate this process: #!/bin/sh export SAGE_CHECK="yes" # cd to SAGE_ROOT, i.e. the top-level Sage directory for spkg in `ls spkg/standard/ | grep '\.spkg'`; do ./sage -f "$spkg" done If any test suite fails, then it shouldn't mess up your local Sage installation. You would still have a working Sage installation. -- Regards Minh Van Nguyen -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
