I have modified the configuration checks for NaN et al.
Felix now provides isnan, isinf, and isfinite for floating types.
THIS NEEDS TESTING ON NON LINUX PLATFORMS!
--
John Skaller
Felix, successor to C++: http://felix.sf.net
On Mon, 2007-09-10 at 03:19 +1000, skaller wrote:
> I have modified the configuration checks for NaN et al.
> Felix now provides isnan, isinf, and isfinite for floating types.
>
> THIS NEEDS TESTING ON NON LINUX PLATFORMS!
This program:
///
#import
println 1
On 9/9/07, skaller <[EMAIL PROTECTED]> wrote:
> On Mon, 2007-09-10 at 03:19 +1000, skaller wrote:
> > I have modified the configuration checks for NaN et al.
> > Felix now provides isnan, isinf, and isfinite for floating types.
> >
> > THIS NEEDS TESTING ON NON LINUX PLATFORMS!
It's almost working
This is odd. I can't directly call the isinf function:
#import
val foo = Double::isinf;
Results in:
CLIENT ERROR
Can't find isinf
In ./foo.flx: line 2, cols 1 to 24
2: val foo = Double::isinf;
What's going on?
--
On 9/9/07, Erick Tryzelaar <[EMAIL PROTECTED]> wrote:
> This is odd. I can't directly call the isinf function:
Looking at the debug output, it looks like it's not even using the
right function. It references this:
instance [T: TYPE where (T isin (short || ushort || int || uint ||
long || ulong ||
This is very strange, and I think possibly another inlining bug. If I run this:
#import
val fzero = 0.0f;
println $ (1.0f/fzero);
It returns "inf". But if I use the mk pipeline:
> bin/flxg -Ilib std foo
> g++ -fPIC -bundle -c -fno-common -Wall -Wno-invalid-offsetof -Wfatal-errors
> -g -O3 -fom
On Sun, 2007-09-09 at 13:40 -0700, Erick Tryzelaar wrote:
> This is odd. I can't directly call the isinf function:
>
> #import
> val foo = Double::isinf;
>
> Results in:
>
> CLIENT ERROR
> Can't find isinf
> In ./foo.flx: line 2, cols 1 to 24
> 2: val foo = Double::isinf;
>*
On Sun, 2007-09-09 at 14:15 -0700, Erick Tryzelaar wrote:
> On 9/9/07, Erick Tryzelaar <[EMAIL PROTECTED]> wrote:
> > This is odd. I can't directly call the isinf function:
>
> Looking at the debug output, it looks like it's not even using the
> right function. It references this:
>
> instance [T
On 9/9/07, skaller <[EMAIL PROTECTED]> wrote:
> Delete all the *.par files. Felix parser auto-dependency checking
> only works for include files not #include or #import.
I tried that, and actually did a completely fresh rebuild, and I still
was getting the error.
-
On Sun, 2007-09-09 at 14:48 -0700, Erick Tryzelaar wrote:
> This is very strange, and I think possibly another inlining bug. If I run
> this:
>
> #import
> val fzero = 0.0f;
> println $ (1.0f/fzero);
>
> It returns "inf". But if I use the mk pipeline:
>
> > bin/flxg -Ilib std foo
> > g++ -fPIC
On 9/9/07, skaller <[EMAIL PROTECTED]> wrote:
> > It returns "+inf". All very mysterious. Are you getting this on your end?
>
> Nope, I get:
>
> [EMAIL PROTECTED]:/work/felix/svn/felix/felix/trunk$ f er2
> +inf
>
> where my 'f' bash command uses --force to ensure everything
> is recompiled
On Sun, 2007-09-09 at 19:10 -0700, Erick Tryzelaar wrote:
> Found it! My mistake. I got something similar, a file called "flx"
> that's just this:
>
> #!/bin/sh
> flxdir=/Users/Shared/erickt/Projects/felix/build.git
> $flxdir/bin/flx --test=$flxdir $*
>
> *The* problem was that I was working in
On 9/9/07, skaller <[EMAIL PROTECTED]> wrote:
> Wouldn't that use the wrong binaries too? That is *supposed*
> to delegate EVERYTHING to the $flxdir directory.
> [i.e. completely isolate each distinct Felix installation]
Yep, it was all wrong :( But, now it's working again :)
13 matches
Mail list logo
