nanogui: A couple of patches and a couple of other notes
Subject:
Re: A couple of patches and a couple of other notes
From:
"Greg Haerr" ####@####.####
Date:
5 Apr 2000 01:35:12 -0000
Message-Id: <0e5101bf9e9d$c014b5a0$15320cd0@gregh>
but there are a couple that
: seem to be just bugs in the mwin engine/build process:
[snip]
Steven -
Great job with the fixes, and their diff'd with pre6 to boot!
I'll add all of them immediately, they look fine.
Are you using Microwindows for DOS? I was wondering
if anyone used that non-select method of polling for input...
In regards to the printf/perror thing, Al made a suggestion
about a month ago that perhaps we eliminate printf()
entirely for systems that are small or don't have anywhere
to printf to. I'm trying to think of the proper config
option for both perror, printf, fprintf, and that sort of thing.
I commented that perhaps a config option to not use
libc stdio might be good. (We already have HAVE_FILEIO_SUPPORT
to remove fopen/fclose etc.)
There are a
: couple of other things that we haven't tried to patch yet that also need
: to be fixed, though; the most notable is that FLOAT is defined to be
: float in windef.h (which is good; this is preferrable to defining it as
: double), but the floating-point constants in graph3d.h (pi and epsilon)
: will default to being doubles (they should be suffixed with an 'f' to make
: it clear that they're float constants) and cos(), sin(), etc. take double
: arguments, not float, so they should be changed to cosf(), sinf(), sqrtf()
: and atanf(). (Also, the 'atan' call in angle() should probably be an 'atan2',
I'm open for input as to whether the math functions should all be defined
for FLOAT or DOUBLE, or either. I thought most math libs
converted float to double and performed all math in double precision,
so that actually it was MORE costly to use floats. Also, is
cosf() etc available in all math libs?
Regards,
Greg
