Re: Problems in 'make bench' under Cygwin/64 with testing elliptic

[John Cremona <john.cremona@gmail.com>]

As far as I know, the people porting Sage to Cygwin have everything
(including pari) working except perhaps Atlas, so it should be no
surprise that pari builds ok under Cygwin.  I can ask for details on
sage-devel if desired.

John

On 5 January 2015 at 13:53, Andreas Kellers <andreas.kellers@gmail.com> wrote:
> PARI version is 2.7.2 (stable), gcc is 4.8.3.
>
> Building GP with debug mode, I get the same error.
>
> Next, I'll try to provide a backtrace.
>
> On Mon, Jan 5, 2015 at 2:18 PM, Bill Allombert
> <Bill.Allombert@math.u-bordeaux.fr> wrote:
>>
>> On Mon, Jan 05, 2015 at 10:09:56AM +0100, Andreas Kellers wrote:
>> > I'm able to build GP using Cygwin/64 under Win 7 (using gmp and current
>> > GP
>> > extra packages), but running 'make bench' states problems while testing
>> > elliptic (diff files attached).
>> >
>> > I also tried a clean build without extra packages with the same result.
>> >
>> > Any hints? If necessary I can provide further information about my
>> > environment.
>>
>> We never tested Cygwin/64. If I read the documentation correctly,
>> Cygwin/64 use
>> the LP64 model (as on linux) rather than the usual LLP64 model on windows
>> which
>> is not supported by PARI. So you should be able to build 64bit PARI
>> binaries
>> that will run on windows which is great news!
>>
>> This is probably a compiler related issue. Either the compiler generates
>> wrong code
>> or the PARI code is not standard compliant at some point.
>>
>> If you use './Configure -g' for debug mode, do you get the same error ?
>>
>> If you have a debugger, you can try to provide a backtrace by setting a
>> breakpoint
>> on pari_err_INV() or pari_err().
>>
>> Which version of PARI are you building ?
>> Which version of gcc are you using ?
>>
>> You can check here the list of systems and compilers we test:
>> <http://pari.math.u-bordeaux.fr/buildlog.html>
>>
>> Cheers,
>> Bill.
>>
>