Ingo, Do you feel like having a go at this on pdla-core instead? My offer of copying your branch over to your fork of pdla-core stands.
-----Original Message----- From: Ingo Schmid Sent: Thursday, April 25, 2019 8:51 PM To: pdl-devel@lists.sourceforge.net Subject: Re: [Pdl-devel] hidden type conversion Thank you Luis, I think I have a different issue, though, I think. I've got a piddle with c type complex double, pdl type cdouble. There's no '<' or '>' for that c type, ops.pd takes care that no overloaded code is generated, otherwise that fails. Yet when I say $a=ones(3)+ci; $a=xvals(3)+ci; print $a<$b; I get as result the comparison of the real parts. I suspect some 'clever' PDL_ANYVAL logic is going on here, yet I don't get it. Best Ingo PS: Check out the nat_complex branch https://github.com/fantasma13/pdl.git On 25.04.19 20:38, Luis Mochan wrote: > I made a similar discovery and asked on the pdl-general list on > Jan. 18. The answer I got from Derek Lamb is that the issue has been > corrected on the complex_atan2 branch in github. I recall there have been > several other changes to PDL::Complex on that branch. > > Regards, > Luis > > > On Thu, Apr 25, 2019 at 08:01:01PM +0200, Ingo Schmid wrote: >> Hi, >> >> comparison of complex numbers make sense only to some extent, < , > are >> not defined. Yet even though in ops.pd they are disabled, somehow the >> real part gets compared. I guess there's a hidden conversion to double. >> I just can's see where that's happening. >> >> Does anyone know? >> >> Best wishes >> >> Ingo >> >> >> > >> _______________________________________________ >> pdl-devel mailing list >> pdl-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/pdl-devel > _______________________________________________ pdl-devel mailing list pdl-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pdl-devel _______________________________________________ pdl-devel mailing list pdl-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pdl-devel
