Re: Some bugs?

[Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr>]

On Fri, Jul 03, 2009 at 04:45:28PM +0100, Jason Moxham wrote:
>
>> This seems to work on linux32 though. Could you compare the output of
>> polchebyshev(n,2) (for n=26..50) on cygwin and linux ?
>>
>
> Typical,  one of the machines I use for linux32 has no bison , the other  

You can install bison if you need it.

> one's network card has broken , and my main one is only booting the wrong 
> kernel :( , so here is a cygwin with no gmp which appears to work...

>                 GP/PARI CALCULATOR Version 2.4.2 (development)
>           i686 running cygwin (ix86/GMP-4.2.1 kernel) 32-bit version
>
> parisize = 4000000, primelimit = 500000
> ? polchebyshev(27,2)
> %2 = 134217728*x^27 - 872415232*x^25 + 2516582400*x^23 - 4244635648*x^21 
> + 46425
> 70240*x^19 - 3218161810*x^17 + 1657840932*x^15 - 592086047*x^13 +  
> 144320973*x^11
> - 23209513*x^9 + 2320951*x^7 - 130320*x^5 + 3393*x^3 - 26*x

>                 GP/PARI CALCULATOR Version 2.4.2 (development)
>                i686 running cygwin (ix86 kernel) 32-bit version
> ? polchebyshev(27,2)
> %2 = 134217728*x^27 - 872415232*x^25 + 2516582400*x^23 - 4244635648*x^21 
> + 46425
> 70240*x^19 - 3451650048*x^17 + 1778122752*x^15 - 635043840*x^13 +  
> 154791936*x^11
> - 24893440*x^9 + 2489344*x^7 - 139776*x^5 + 3640*x^3 - 28*x

Theses results are different (the second is correct)
However I see that revision 9957 fixed a memory corruption bug in
polchebyshev(,2). This might explain the discrepancy.
If you want to be sure apply the patch given by 'svn diff -c 9957'.
But it is more important to focus on the SVN HEAD branch, because 
2.4.2 is not supported.

Cheers,
Bill.