At Sat, 31 Oct 2015 23:14:11 +0300, Dmitry Pavlov wrote:
> I vaguely remember that extflonum support was supposed to be turned
> off by default in 32-bit Linux releases. OK. But current 32-bit Windows
> nightly builds do not provide it either. I did not notice when they
> stopped to provide it.

When I try the 32-bit Windows builds, they seem to have extflonums enabled.

Are you using the pre-built versions or building your own? If you build
your own, note that extflonums will not become available until the
"racket-lib" package is installed, since that will install the needed
DLL.

Otherwise, can you say exactly which build you're using and how you're
checking for extflonums? Maybe that will help me see which step was
different or wrong for me. I tried (or intended to try!) the "32-bit
x86" links under "Minimal Racket" from both the Utah and Nortwestern
sites.

JFYI, that the Northwestern builds are created using MinGW on OS X,
while the Utah builds are created on Windows with MSVC. The release
candidate for v6.3 (which also seems to work for me) is created with
MinGW on Linux.


> Moreover: I just built Racket from github source on 32-bit Linux with
> --enable-extflonum switch to ./configure, and the build version
> somehow misses the extflonums:
> 
>   extfl+: unsupported on this platform
> 
> I suppose that maybe something went wrong in the build scripts
> and/or macros.

I guess the `--enable-extflonum` flag is misleading. It's enabled by
default, but a build includes extflonum support only if compatible
compiler flags are provided. You probably need to provide

  CPPFLAGS="-msse2 -mfpmath=sse"

to `configure` to enable extflonums on 32-bit Linux.

I will try to make `configure` complain when `--enable-extflonum` is
specified and the compiler does not provide enough support for
extflonums.

-- 
You received this message because you are subscribed to the Google Groups 
"Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to racket-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to