Subject: Re: [ublas] [bindings] Lapack generator
From: Thomas Klimpel (Thomas.Klimpel_at_[hidden])
Date: 2009-01-11 06:02:49
Rutger ter Borg wrote:
> Thomas Klimpel wrote:
> > - the fortran INTEGER type is translated to int instead of integer_t
> Where is the point where ints are translated to integer_t? Is the needed
> code path identical to that of fcomplex_t, traits::complex_f*, or is this
> case different?
It's not, because everything is a pointer at that point, but the conversion cannot be done with reinterpret_cast<integer_t*>(...). The conversion is done "as soon as possible", which sometimes means "not at all" (an integer workspace array can't be converted, for example). The "info" argument is converted to int on the user side, the other arguments are converted to std::ptrdiff_t on the user side, if possible.