[sage-devel] Re: Request for testing: FLINT-1.01.spkg
On Dec 9, 4:59 pm, mabshoff [EMAIL PROTECTED] dortmund.de wrote: Hi Bill, I added FLINT-1.01 to Sage-2.9.alpha2. The spkg can be found at http://sage.math.washington.edu/home/mabshoff/flint-1.01.spkg While testing on sage.math I got the following: Testing _fmpz_poly_scalar_mul_si()... ok Testing fmpz_poly_scalar_mul_si()... ok Testing _fmpz_poly_scalar_mul_fmpz()... ok Testing fmpz_poly_scalar_mul_fmpz()... ./spkg-check: line 16: 94786 Floating point exception./fmpz_poly-test real4m39.394s user4m44.254s sys 0m6.256s sage: An error occurred while installing flint-1.01 Sorry, this was on bsd, i.e. Dual CoreQuad OSX 10.5.1. It is fine on sage.math, sorry for the noise. FLINT 1.0 worked fine on that box. Cheers, Michael I ran the testsuite once before that without failure, but this time I compiled with make=MAKE -j4. Any idea what could be wrong? Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
mabshoff wrote: Hi Bill, I added FLINT-1.01 to Sage-2.9.alpha2. The spkg can be found at http://sage.math.washington.edu/home/mabshoff/flint-1.01.spkg On Fedora 7 32 bits: Testing _fmpz_poly_scalar_mul_fmpz()... ok Testing fmpz_poly_scalar_mul_fmpz()... ./spkg-check: line 16: 10706 Floating point exception./fmpz_poly-test real7m42.054s user7m39.012s sys 0m1.923s sage: An error occurred while installing flint-1.01 Jaap --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
Hi Michael, is that also a 32 bit box? Looks like another bug surfacing. Again it is probably a bug in the test code itself. A divide by zero in some corner case. I'll try to dig it out. Bill. On 9 Dec, 16:08, mabshoff [EMAIL PROTECTED] dortmund.de wrote: On Dec 9, 4:59 pm, mabshoff [EMAIL PROTECTED] dortmund.de wrote: Hi Bill, I added FLINT-1.01 to Sage-2.9.alpha2. The spkg can be found at http://sage.math.washington.edu/home/mabshoff/flint-1.01.spkg While testing on sage.math I got the following: Testing _fmpz_poly_scalar_mul_si()... ok Testing fmpz_poly_scalar_mul_si()... ok Testing _fmpz_poly_scalar_mul_fmpz()... ok Testing fmpz_poly_scalar_mul_fmpz()... ./spkg-check: line 16: 94786 Floating point exception./fmpz_poly-test real4m39.394s user4m44.254s sys 0m6.256s sage: An error occurred while installing flint-1.01 Sorry, this was on bsd, i.e. Dual CoreQuad OSX 10.5.1. It is fine on sage.math, sorry for the noise. FLINT 1.0 worked fine on that box. Cheers, Michael I ran the testsuite once before that without failure, but this time I compiled with make=MAKE -j4. Any idea what could be wrong? Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
I can replicate this bug on a 32 bit box I have access to but not on any 64 bit box. So happy days. I'll try and issue a patch once I figure out the problem. It isn't jumping out at me at present. Bill. On 9 Dec, 17:11, Bill Hart [EMAIL PROTECTED] wrote: Hi Michael, is that also a 32 bit box? Looks like another bug surfacing. Again it is probably a bug in the test code itself. A divide by zero in some corner case. I'll try to dig it out. Bill. On 9 Dec, 16:08, mabshoff [EMAIL PROTECTED] dortmund.de wrote: On Dec 9, 4:59 pm, mabshoff [EMAIL PROTECTED] dortmund.de wrote: Hi Bill, I added FLINT-1.01 to Sage-2.9.alpha2. The spkg can be found at http://sage.math.washington.edu/home/mabshoff/flint-1.01.spkg While testing on sage.math I got the following: Testing _fmpz_poly_scalar_mul_si()... ok Testing fmpz_poly_scalar_mul_si()... ok Testing _fmpz_poly_scalar_mul_fmpz()... ok Testing fmpz_poly_scalar_mul_fmpz()... ./spkg-check: line 16: 94786 Floating point exception./fmpz_poly-test real4m39.394s user4m44.254s sys 0m6.256s sage: An error occurred while installing flint-1.01 Sorry, this was on bsd, i.e. Dual CoreQuad OSX 10.5.1. It is fine on sage.math, sorry for the noise. FLINT 1.0 worked fine on that box. Cheers, Michael I ran the testsuite once before that without failure, but this time I compiled with make=MAKE -j4. Any idea what could be wrong? Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
I know where the bug is. It is in our integer multiplication code. There's a short piece of code there which is really dodgy and needs rewriting since it has faulted a number of times before. It's been on my todo list for about a week. As soon as I've fixed it I'll issue an update. But it is going to take a few hours. Bill. On 9 Dec, 17:40, Bill Hart [EMAIL PROTECTED] wrote: I can replicate this bug on a 32 bit box I have access to but not on any 64 bit box. So happy days. I'll try and issue a patch once I figure out the problem. It isn't jumping out at me at present. Bill. On 9 Dec, 17:11, Bill Hart [EMAIL PROTECTED] wrote: Hi Michael, is that also a 32 bit box? Looks like another bug surfacing. Again it is probably a bug in the test code itself. A divide by zero in some corner case. I'll try to dig it out. Bill. On 9 Dec, 16:08, mabshoff [EMAIL PROTECTED] dortmund.de wrote: On Dec 9, 4:59 pm, mabshoff [EMAIL PROTECTED] dortmund.de wrote: Hi Bill, I added FLINT-1.01 to Sage-2.9.alpha2. The spkg can be found at http://sage.math.washington.edu/home/mabshoff/flint-1.01.spkg While testing on sage.math I got the following: Testing _fmpz_poly_scalar_mul_si()... ok Testing fmpz_poly_scalar_mul_si()... ok Testing _fmpz_poly_scalar_mul_fmpz()... ok Testing fmpz_poly_scalar_mul_fmpz()... ./spkg-check: line 16: 94786 Floating point exception./fmpz_poly-test real4m39.394s user4m44.254s sys 0m6.256s sage: An error occurred while installing flint-1.01 Sorry, this was on bsd, i.e. Dual CoreQuad OSX 10.5.1. It is fine on sage.math, sorry for the noise. FLINT 1.0 worked fine on that box. Cheers, Michael I ran the testsuite once before that without failure, but this time I compiled with make=MAKE -j4. Any idea what could be wrong? Cheers, Michael --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
On Dec 9, 6:53 pm, Bill Hart [EMAIL PROTECTED] wrote: I know where the bug is. It is in our integer multiplication code. There's a short piece of code there which is really dodgy and needs rewriting since it has faulted a number of times before. It's been on my todo list for about a week. As soon as I've fixed it I'll issue an update. But it is going to take a few hours. Thanks, Bill Michael SNIp --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
[sage-devel] Re: Request for testing: FLINT-1.01.spkg
I have issued a service release, FLINT 1.0.2 to be found at http://www.flintlib.org/ which should fix this issue. It was a divide by zero which occurred in a corner case in the tuning code for the FLINT integer multiplication code. The code which is there now is much neater, much more robust, and of course more correct than what was there before. I've tested this release on 32 and 64 bit x86. Hopefully we've run out of bugs now. Once again, thanks for the testing feedback. Bill. On 9 Dec, 17:57, mabshoff [EMAIL PROTECTED] dortmund.de wrote: On Dec 9, 6:53 pm, Bill Hart [EMAIL PROTECTED] wrote: I know where the bug is. It is in our integer multiplication code. There's a short piece of code there which is really dodgy and needs rewriting since it has faulted a number of times before. It's been on my todo list for about a week. As soon as I've fixed it I'll issue an update. But it is going to take a few hours. Thanks, Bill Michael SNIp --~--~-~--~~~---~--~~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~--~~~~--~~--~--~---
