On Thu, 14 Jun 2007, Gerrit Bruchhäuser wrote:

> If it were just the single one error I could easily ignore it. But the total
> amount is by far too impressive.

I know. And if you google for 'openssl valgrind' and similar combinations, you
can find others in the same situation. It seems you can rebuild openssl with
-DPURIFY to fix at least some of them.

Clearly this is because they use uninitalized memory to feed the randomizer
with, and in that case uninitialized memory isn't a bad thing...

> If OpenSSL is of that quality, is there an alternative which I could
> fallback with libcurl?

I don't think it is a sign of bad quality. But libcurl can also be built with
GnuTLS, NSS or even yassl so the choice is yours!

-- 
  Commercial curl and libcurl Technical Support: http://haxx.se/curl.html