So the main problems start to crop up when you start stripping out assumptions from the constructor of basic sympy objects. This breaks compatibility, and very quickly the number of errors goes up. You can view the natural progression of what you suggest on this branch: - http://github.com/haz/sympy/commits/disconnect-assumptions-2
<http://github.com/haz/sympy/commits/disconnect-assumptions-2> Up to this commit, errors are fixed up and things go from 293 passed, 17 failed, 20 exceptions down to 300 passed, 22 failed, 8 exceptions: - http://github.com/haz/sympy/commit/d25877c0998bda73033a16b30b12494c53da6eec <http://github.com/haz/sympy/commit/d25877c0998bda73033a16b30b12494c53da6eec>-- Note, even though that last commit removes an entire test suite, it only saves you a single exception -- the other commits contribute a lot. Now when you start to remove the assumptions argument to the constructor of Symbol, via this commit, things start to rear their ugly head: 291 pass, 11 fail, 27 exceptions: - http://github.com/haz/sympy/commit/f8e4e04f6f92b54f777f624a8840ed5dd3571b8c <http://github.com/haz/sympy/commit/f8e4e04f6f92b54f777f624a8840ed5dd3571b8c> So I think the reason the first attempt looks so bad (77 exceptions), is because I've gone deeper into removing the assumptions from the constructors of various sympy objects. I would expect to see the same when I continue down this second attempt (that is without the predicate stuff, the Application class, etc). Hopefully this sheds some light on the issues I'm facing. Feedback on the commits in dissconnect-assumptions-2 would be great to have. Cheers Christian On Tue, Jun 15, 2010 at 10:15 PM, Ondrej Certik <ond...@certik.cz> wrote: > Hi, > > I've tried the disconnect-assumptions branch and I got some recursive > exception problems, so I did > > git checkout HEAD~2 > > and then they disappeared, but I still got lots of other exceptions > and failures. Run the test using: > > Then this branch here: > > http://github.com/certik/sympy/tree/assum > > gives: > > $ bin/test sympy/core/ > [...] > tests finished: 293 passed, 17 failed, 5 expected to fail, > 1 expected to fail but passed, 20 exceptions, in 1.55 seconds > > e.g. totall 17+20=37 failures, I think. > > but the disconnect-assumptions branch (without the last two commits) gives: > > > $ bin/test sympy/core/ > [...] > tests finished: 249 passed, 6 failed, 6 expected to fail, 77 exceptions, > in 4.73 seconds > > e.g. total 6+77=83 failures. Also the number of passed tests is lower. > > > So I am a bit confused. I think the goal would be to get my original > branch to pass (e.g. fix the rest 37 failing tests -- resp. the > problem why they fail) and then run the full test suite, fix the rest. > Then the assumptions should be gone. Then we can start polishing it, > and start integrating some new cool stuff from Ronan and other guys. > > Is there any reason why the above approach doesn't work? I think in > the disconnect-assumptions, there are too many new patches and things > and it's hard to debug it, when something fails (as it always > something fails). > > Ondrej > -- You received this message because you are subscribed to the Google Groups "sympy-patches" group. To post to this group, send email to sympy-patc...@googlegroups.com. To unsubscribe from this group, send email to sympy-patches+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/sympy-patches?hl=en.
