Mailing List Archive: GNUPG and a RNG

Greetings...

I just finished catching up on all the emails in the listserv archives. :)
Anyway, it would seem that with all this debate about RNGs, that perhaps it
would be a good idea to make the RNG in GNUPG a little more modular. Like
someone else pointed out, it would be nice to be able to run GNUPG on
systems without the /dev/random device.

One other question... I noticed as I was thrashing through the code
yesterday (in my failed attempt at cross-compiling) that the only other
major stumbling block for cross-compiling is some heavy usage of signals in
the cipher library (if I understood the code correctly). Could someone give
me a feel for what these are being used for? :)

Peter, is there a URL where I can find your paper on the practically strong
random numbers?

Thanks.

D.

Dave Smith <dave@raystewart.com> writes:

> One other question... I noticed as I was thrashing through the code
> yesterday (in my failed attempt at cross-compiling) that the only other
> major stumbling block for cross-compiling is some heavy usage of signals in
> the cipher library (if I understood the code correctly). Could someone give

Signals ???

GNUPG does only use SIGUSR1 if the option --run-as-shm-coprocess is
used (to trigger a check whether data from the frontend is available).

Disable this for Windows (configure should already do this) or use
Windows "shared memory". I used to code something like this with
a simple file and filelocking (yes - polling).

Other signals are only catched

> Peter, is there a URL where I can find your paper on the practically strong
> random numbers?

See Peter's homepage.

Werner