> -----Original Message-----
> From: Ted Powell [mailto:"Ted Powell <ted"@tgivan.com]
> Sent: Friday, April 28, 2000 5:59 PM
> When I ran openssl-0.9.5a's "make test", the tmp.bntest file that was
> written contained (out of 1800+ tests) seven expressions
> which Solaris 7's 'bc' evaluated to give non-zero results.
> When I feed the problematic tmp.bntest to GNU bc
> all the expressions evaluate to zero, as they should.
I had similar results on AIX 3.2.5 with SSLeay: the native bc produced
errors, Gnu bc passed the files. A little testing revealed a bona fide
error in the native bc (fixed in AIX 4.2.1). Unfortunately, it's been a
while since I ran across this, and I don't remember what I used to verify
that the error existed.
Of course, it may not be a bug per se. I don't recall what statements, if
any, the AIX or Solaris bc man pages make about precision and scale limits.
(Yes, bc is supposed to be "arbitrary precision", but obviously it has
finite limits, since the machine it's running on is finite. I suppose a bc
implementation could impose limits too strict to pass all the OpenSSL BN
tests.) Do Unix95 or any of the other specs require a bc implementation
sufficient to pass all the BN tests? I don't know.
Michael Wojcik [EMAIL PROTECTED]
MERANT
Department of English, Miami University
______________________________________________________________________
OpenSSL Project http://www.openssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
