The patch looks good to me.
David Lee wrote:
While chasing another problem, I stumbled across some uncleanliness in
"replace/setenv.c" and "replace/unsetenv.c". (This probably doesn't
affect Linux.)
1. They weren't using "<portability.h>", but instead re-declared things
themselves. I have tidied this.
2. One of the declarations of "unsetenv" had "int" return type. Linux
and
BSD seem to use "void", so I have corrected to this. (Curiously, when
Solaris introduced setenv/unsetenv (previously absent) in the
relatively recent version 10, they made it "int unsetenv(...)". But
I'm happy to stick with "void" if that's what folk want.
3. From "replace/setenv.c", the return value of its internal "putenv()"
wasn't being conveyed back to the caller. I have corrected this.
<digression>
This last point raises the wider question of the heartbeat not
checking the "setenv" return code. Now that the software is so
reliant on "setenv" it would seem wise (probably after the forthcoming
release) for us to begin to address this: an error-return, though
rare, would probably be serious.
good point. I will do that.
-Guochun
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
