> Date: Mon, 4 Sep 2000 12:54:49 -0500 (EST)
> From: Dan Moore <[EMAIL PROTECTED]>
>
> however, the O_BINARY block in lib/system.h that tries to include <io.h>
> still doesn't work. the problem there is that just because O_BINARY is
> defined does not necessarily imply that <io.h> exists (it doesn't on
> BeOS). also, in that same block of conditioned defines, the second set
> (the !O_BINARY block) of defines is what should be used on BeOS.
>
> so there are two ways to address all the O_BINARY related code/defines to
> allow it to run on BeOS. one is to simply check to see if it is being
> compiled on __BEOS__ and then undef O_BINARY away (see the gg
> lib/system.h).
I think it is safer to say
#if O_BINARY && !defined __BEOS__
instead of "#if O_BINARY". In my experience, undefining
system-defined symbols might be dangerous. Will the above work for
BeOS?
Karl, could you comment on this?
