Hi David, > Although it is true that not everyone can run tests against commercial > software, I would have thought a significant proportion of Sage users > could. There is already an interface to Mathematica. Many Sage users and > developers work in universities, which often have Mathematica licenses.
[...] > IMHO, it is better if the person writing the test is not the same person > who wrote the code being tested. Sometimes one makes an assumption about > how a function should work, where the developer(s) make a different > assumption. Neither are wrong, but it highlights areas where perhaps > documentation should be clarified. > > My experience in R+D has often showed me that comparing two totally > different methods is useful. > > Anyway, it seems my view is a minority one here. If this is a call for a vote ;-), let me tell that I completely agree with the point of view that in an ideal world, tests should be written *before* the code and by a *different* person (extreme/peer programming). I'm proud that we have done this several time together with Nicolas. My experience is that the resulting code+tests are undoubtedly better. However, in my personal experience, comparing with other software is rarely doable in practice. But this is certainly due to the kind of computations I need. Instead of this, I strongly relies on coherency test and test against proved theorems. Ideally, the proof of those theorem should be checked by a computer and thus can be seen as a concurrent software. Cheers, Florent -- 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
