[AGENT++] [PATCH] snmp++: (several simple patches)

[Jochen Katz]

Hi,

>   The small fix for the initial rid calculation I reported
>   earlier.

was already done.

>   Fixes initialization of the salt on systems where integers are
>   not 32 bit, i.e. 64 bit systems.

Funny: I had a discussion with a colleague about integer sizes this
week. GCC/VC on 64 bit Linux/Windows use a 32 bit integer. And VC on 64
bit Windows has a 32 bit long for "compatibility reasons" ;-)
I thought that I have seen a system that uses a 64 bit int, but I don't
know it any more.
As your code looks better than the existing: accepted fro teh next release.

>   The "clean" target also cleans core files like core.12345.
>   Slightly simplifies the dependencies to the target "all"

Accepted.

>   Makefile rules to build and install shared and static libraries
>   separately.

Accepted. The "ifneq ($(wildcard..." is removed, as it does not have any
effect (even before your change).

>   Add source code documentation for start_poll_thread() and
>   stop_poll_thread().

Accepted, but slightly modified.

>   Don't call pthread_exit() at end of thread execution.  This is
>   actually not necessary; the old code was 100% correct.  The only
>   effect of this change is that if pthread_exit() is never called,
>   glibc does not open some shared library which valgrind reports
>   as still reachable memory upon exit.  If the thread is simply
>   allowed to reach the end of its run() function, that library is
>   not opened.

Hmm, I think we kee the pthread_exit() for now.

Thanks and Regards,
  Jochen