Bug#719015: Update: the magic number should be 1.9999999403953553

2013-08-09 Thread Martin Brain
Sorry. The previous examples hit a discrepancy between strtof and strtod. The nearest float to 1.99940395356f should be 0x1.0p+1f, rather than 0x1.fep+0f. If instead the number 1.999403953553f is used (see updated test case), then it correctly exposes the double rounding bug. 1.

Re: Location for Fortran90 module files

2013-08-09 Thread Michael Banck
On Fri, Jul 26, 2013 at 02:37:15PM +0200, Matthias Klose wrote: > Am 25.07.2013 14:25, schrieb Michael Banck: > > What to do about the compatibilty is a different matter, did this come > > up before? > > looks like one hand in the debian-science team doesn't know what the other > hands > are doin