On Jun 4, 11:24 am, Robert Bradshaw <rober...@math.washington.edu> wrote:
> > Maybe you're claiming that Sage offers no advantage over completely > closed systems or manually managing fragmented, hard to configure, > specialized libraries, but I think both are (huge) steps in a good > direction. I made no such claim. Sage literature itself compares the Sage "system" explicitly to those systems to which it is supposed to provide an alternative. Mathematica, Maple... There are such systems that are open source, including Axiom / Fricas, Maxima (and Matlab clones too). Loading and running Sage is apparently a challenge for some people. I do not have any experience trying that myself, nor do I have recent experience downloading any Axiom variant. I do, however, have recent experience downloading Maxima on ubuntu, Windows XP, Windows 7 and Mac OS X 10.*. And it is trivial, requiring downloading a file and executing the program in that file. I have also downloaded and installed Mathematica (though not a recent Maple). These are quite simple. So, to me the comparison is not between Sage and some (non-existent) alternative say, bizarro-Sage that tries to do the same thing as Sage. To try to get back to your question (allegation?) Do I think that Sage offers no advantage over X Y Z. No, but I think that for many users it offers no advantage over (say) Maxima. Because (1) Maxima runs on their system. and if that were not enough (2) Maxima does everything they are interested in doing, and (3) Just in case they cared, it is free and open source. While I have no quarrel with people who wish to give access to their source code (I do it myself), I find the need to duplicate effort in order to maintain license purity to be the flip slide of the coin. Instead of duplicating the functionality of proprietary code, the Sage project seems to be requiring people to duplicate the functionality of open-source code!! (but perhaps this quarrel with GMP has been resolved? I have not followed it closely.) > > > when it appears that experts have difficulty even installing already- > > compiled packages? > > > It seems to me that Sage is hard to use, hard to change, apparently > > even hard to install. > > You've never tried to do any of the above. Correct. I am relying on reports from others. I am probably not the Sage target audience (Computer science primarily. Math secondarily. Not much interest in Magma-style stuff.) > Some people find all of the > above easy, and others find it hard, (and then some have to be > instructed to plug in their computer before complaining that it won't > turn on). I'd say Sage is slightly harder to install, (much!) easier > to change, and about equivalent to in terms of usability usability > than any of the commercial offerings. Again, you talk about commercial offerings, as though Sage were alone in the open-source department. But it is not. There are some 52 computer algebra systems listed in the Wikipedia article comparing systems. 33 are free. I'm most familiar with Macsyma/ Maxima, but there are 51 others. Now William Stein will claim that Sage is so much more that a computer algebra system. It does all this other stuff too. Even though it doesn't have to, to be a viable alternative to M,M .... Grasping for an analogy, I am reminded of competitive sales pitches for mini-vans, where one -7-seat vehicle is billed as advantageous because it has 14 cup-holders. Maybe Sage should install more cup-holders on that little car illustration? As for how easy it is to change some system, consider how one could change Maxima. Many changes to the system are easily done with ordinary commands, but say you wanted to alter the underlying lisp code. A silly change would be to make all the coefficient multiplication results in the rational function package result in the number 1. That is rat(x+1)^2 would end up as x^2+x+1 instead of x^2+2*x+1. This could be done by changing the definition of "ctimes". Here is how you could do it. Note that (%i5) is the command line prompt... (%i5) :lisp (defun ctimes(a b) 1) (%i6) That's it. No recompiling, no makefile, no anything. > > > This is not surprising. > > > The claim that one could prove that some subsystem is correct by > > examining its source code is an "in principle" argument of very little > > practical use. Has anyone proved that any result from Maxima is > > correct by examining the source code? Note that as one step you would > > have to prove that the > > Lisp compiler is correct... > > And the compiler, assembler, and chip design. Yes. Even then you'd have to > trust that no cosmic rays affected the compilation or computation. > It's hard to be absolutely sure of anything with computers involved, > but at least you can decide where to put your trust and prove things > from there, under more reasonable axioms, rather than having to take > the top level as a black box. look at the controversy re 4 color thm proof by computer. > > - Robert -- 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
