Re: [PATCH] Use O_BINARY to detect whether to pass "rb" to popen

> From: Filipe Brandenburger <[email protected]>
> Cc: Eli Zaretskii <[email protected]>,
> 	[email protected],
> 	Filipe Brandenburger <[email protected]>
> Date: Wed, 20 May 2015 22:15:46 -0700
> 
> Get rid of freadonly_mode and fwriteonly_mode which were set from a
> define set by a configure AC_RUN_IFELSE check.
> 
> Instead, use the previously already defined FOPEN_READ_BINARY and
> FOPEN_WRITE_BINARY which are set based on O_BINARY from fcntl.h, which
> will be defined to be non-zero on systems where there is a difference
> between text and binary files.
> 
> Remove the now unnecessary configure check for whether popen works with
> binary mode.  That particular check required running a program built by
> configure, which made it hard or impossible to cross-compile sharutils
> without patching the sources.

I don't really care, but isn't that backwards?  I always thought GNU
projects preferred configure tests to manual #ifdef's based on subtle
attributes of the underlying platforms (such as the fact that O_BINARY
is non-zero).

Maybe I'm missing something, but what is the rationale for these
changes?  IOW, what was wrong with the original code?  (Apologies if I
missed some discussion where this was explained.)

(One reason why you may wish to leave that configure test is that many
modern platforms support "rb" and "wb" in calls to 'fopen', but not in
calls to 'popen'.  AFAIR, glibc doesn't support binary modes in
'popen'.)