|Walter Neumann on Wed, 11 Oct 2006 02:13:17 +0200|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
|Re: New experimental GP parser release 2.4.1.bill-1|
make bench fails from me on "Testing elliptic" using the gmp kernel (it is fine with the default kernel). Here is elliptic-dyn.dif:
*** ../src/test/32/elliptic 2006-10-09 15:15:44.000000000 -0400 --- gp.out 2006-10-10 20:00:21.000000000 -0400 *************** *** 158,164 **** ? precision(cmcurve) 38 ? getheap ! [139, 5245] ? print("Total time spent: ",gettime); ! Total time spent: 336 ? \q --- 158,164 ---- ? precision(cmcurve) 38 ? getheap ! [141, 5259] ? print("Total time spent: ",gettime); ! Total time spent: 32 ? \q GP/PARI CALCULATOR Version 2.4.1 (bill) i686 running linux (ix86/GMP-4.1.4 kernel) 32-bit version compiled: Oct 10 2006, gcc-4.1.1 20060525 (Red Hat 4.1.1-1) (readline v5.0 enabled, extended help enabled) --walter On Tue, 10 Oct 2006, Bill Allombert wrote:
Hello PARI-dev, I have made available a new version of my experimental GP parser: <http://pari.math.u-bordeaux.fr/~bill/pari-2.4.1.bill-1.tar.gz> The experimental way to handle local variables has been removed. Instead this version implement local variables and recursives functions the same way as GP. This was the main incompatibility with regular GP. It also fixes a lot of bugs that were in the previous version. In particular, this passes the GP2C test-suite. ( Run make gp;./dotest to try the test-suite.) GP lists are not supported (because I don't know how they are supposed to be managed). On some examples, it is four time faster than standard GP. An experimental feature I plan to add is allowing to use local() (almost) anywhere. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - This new parser is nearly a five-year old project now so it might be time to complete it. But we need your help: If we are going to switch to this parser, we need to build a comprehensive test-suite of GP behaviour. To that end, if you have scripts (or could write some!) that rely more on the language features than on number crushing or on library functions please send them to me together with an example of usage so I can use it as part of a test-suite. Cheers, Bill.