On 12 November 2014 20:35, Ursula Whitcher <whitc...@uwec.edu> wrote:
> Article at > > http://people.uwec.edu/whitchua/notes/sagebugprocess.pdf > > has been updated based on feedback. > > UAW A bit more feedback - from a non-mathematician. 1) It would be better if rather than over-writing an old version of your document, you wrote a revised version and called is sagebugprocess-vesion-2 or similar. 2) Change "Working mathematicians have an alternative to comparing the output of di erent(sic) black boxes: " to "Working mathematicians have a much better alternative than to compare the output of different black boxes, which may well share some of the same code" 2) The phase "A very con dent(sic) user might move straight to the next step: creating a ticket" is wrong. You need to be a particularly confident use to report a bug in a trac ticket. I have reported bugs in software I know very little about, but enough to know there is a bug. Actually fixing the bug requires *ability*, rather than confidence. A totally incompetent user could try to fix the bug, but make a complete mess of it. But you stated something quite different. 3) You have totally missed the point I made, which Karl-Dieter Crisman wrote was a very good point. This is that comparing the output of two black boxes may well be comparing the output of two *identical* bits of code. I just checked the libraries for Mathematica version 7.0.1 on Solaris x86, and a view of the names of them libML64i3.so libML32i3.so liblapack.a libcblas.a libsndfile.so.1 libf77blas.a libatlas.a libgmp.so.3 libQtXml.so.4 libQtGui.so.4 libQtSql.so.4 libQtNetwork.so.4 Qt-Plugins libGLU.so libGLU.so.1 Mesa libQtSvg.so.4 libaspell.so.1 libQt3Support.so.4 libMesaGL.so libMesaGLU.so libQtCore.so.4 would suggest to me that Mathematica is using at least the following two maths libraries * ATLAS * GMP It would be worth checking a more recent release, to see if things have changed, but it is largely irrelevant. The fact remains WRI did use both those maths libraries is good enough. There's no reason Maple could not use them too. GMP is licensed under the GNU Lesser General Public License (LGPL), so Wolfram Research and Maplesoft are well within their rights to do that. In fact, they are well within their rights to hide that fact, as they could just compile GMP into a library libfoo.so, and you would be none the wiser. The mere fact they are "black boxes" means you don't have a clue. If there was a bug in ATLAS, comparing the output of Mathematica and Maple could lead you to a false sense of security, believing you are using two bits of software to perform a computation, but instead using the same bit of code twice. Both you, and the trio of mathematicians seem to have missed this point. This non-mathematician (i.e. me) thinks it is an important point !! 4) It might be worth briefly stating that if (hypothetically) such a bug was found in Sage, rather than just report the bug, the trio could have inspected Sage, determined the code used to find the determinate used algorithms A, B and C, using software packages X, Y and Z, then looked to find the bug in A, B, C or some code which links them together. That would hint Sage gets it right, but be more constructive than just saying it gets it right. You could put that in a few sentances - there is no need to detail the exact process they would have to use. Dave -- 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 http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
