Re: [Openmcl-devel] Re: OpenMCL 1.0-rc1

On Mon, 26 Sep 2005, Luis Oliveira wrote:

> Gary Byers <[email protected]> writes:
>> I realize that it may be something of a PITA to download and install
>> these largish archives (and to repeat the process if there are other
>> "release candidates" between now and the final 1.0), but it'd be
>> helpful if anyone who's able to do so could do so.
>
> Hello,
>
> I just downloaded the release candidate OpenMCL Version 1.0-rc1-050922
> (DarwinPPC32) to check out if the callback bugs I previously reported[1]
> were fixed. Apparently they are but there's a new one. Using the same
> example as before:
>
> ? (external-call "_expect_double" :address return-double)
> f(42.0) => 42.000000
> Lisp side got: 42.0D0
> NIL
> ? (external-call "_expect_float" :address return-float)
> f(42.0f) => 51539608576.678391
> Lisp side got: 42.0
> NIL
>
> Doubles work, and C floats are being converted to Lisp single floats
> correctly, but (and this is a regression, it worked before as you can
> see in my original bug report), the convertion from Lisp single floats
> to C floats is not working as you can see.
>
> Rereading your reply I see that this is probably a known issue already,
> but still...
>
>
> [1] - http://article.gmane.org/gmane.lisp.openmcl.devel/870
>

The double-float 51539608576.678391d0 looks like

#x4228000008015B56

in hex, while the single-float 42.0f0 looks like

#x42280000

Sadly, this is one of those cases where being half right isn't very
helpful (even if it's the upper half.)

[This is now fixed in CVS.]



> -- 
> Luis Oliveira
> luismbo (@) gmail (.) com
> Equipa Portuguesa do Translation Project
> http://www.iro.umontreal.ca/translation/registry.cgi?team=pt
>