Archived
This forum has been archived. Please start a new discussion on GitHub.
NaN became Zero
Hi
I've just found that if my function on returns NaN I receive 0.0 on client.
It was on Ice-3.0.1, Linux 2.4.21-32.0.1.ELsmp i686 i686 i386 GNU/Linux
for both g++ 3.2.3 (server, client) and python 2.4.1 (client)
Do you guys have any plans to fix that?
Well, usually NaN is tough due to lack of portable support. I would say:
give me cmdline option which will enable fenv.h stuff for GNU. The rest
of users could go to hell of passing pair<double, bool>.
Regards,
Andrei
PS: Ice is really, really cool stuff
I've just found that if my function on returns NaN I receive 0.0 on client.
It was on Ice-3.0.1, Linux 2.4.21-32.0.1.ELsmp i686 i686 i386 GNU/Linux
for both g++ 3.2.3 (server, client) and python 2.4.1 (client)
Do you guys have any plans to fix that?
Well, usually NaN is tough due to lack of portable support. I would say:
give me cmdline option which will enable fenv.h stuff for GNU. The rest
of users could go to hell of passing pair<double, bool>.
Regards,
Andrei
PS: Ice is really, really cool stuff

0
Comments
-
NaNs are passing correctly, my mistake
sorry to confuse everybody.
Ice actually doing great job of passing NaNs (which is very important
for my projects). I was confused by FP_NAN from math.h (which is not
NaN repsentation but some enum member). Looks like I need to find
better way to generate NaNs.0