Re: Test failures on V5-8-patches and master

On Thu, Aug 13, 2020 at 03:55:31PM -0700, Wes Hardaker via Net-snmp-coders wrote:
> Magnus Fromreide <[email protected]> writes:
> 
> > I just noticed that T0222snmpv3bulkget_simple depends on SHA authentication
> > and AES privacy but fails to check for this so if one builds the agent
> > without that support the test fails.
> 
> Is it really just that test?  I actually thought that most of the tests
> were using at least sha for authentication.  (no I didn't go look).

That was the only one that FAILED rather than getting skipped when I built
with --with-openssl=internal --disable-privacy.

> In my memory we wrote most of them assuming that snmpv3 support was
> going to be used for most of the tests, as it was the most complex of
> the protocol options.  And few people would turn off v3.

That is a separate issue, you can't turn off v3 even if you want to so it is
always there.

> > Now, I am not certain about exactly what T0222 tests but I think it is about
> > bulkget limits and thus the use of encryption and privacy is just a
> > distraction so I suggest that $PRIVTESTARGS is replaced with $NOAUTHTESTARGS.
> 
> I'd rather see $AUTHTESTARGS, which is in most other places.

Fair enough, I was basing the change on T0220 and T0221 who are both using
NOAUTHTESTARGS.

Looking at T024* seems to suggest that AUTHTESTARGS require some SKIPIF's
though.

/MF