[rbldnsd] pre-announce: rbldnsd-0.990pre0

[Michael Tokarev]

furio ercolessi wrote:
> On Wed, Nov 19, 2003 at 09:56:21PM +0300, Michael Tokarev wrote:
> 
>> - when several RRs of the same type exists in generic dataset,
>>   we now trying to return them in "random" order.  The
>>   "randomization" is very dumb for now.
> 
> It seems to work as advertised :-)
> 
> Thanks! That was quick..
> 
> I think that you did just what Bind does (which is not random at all
> but it fits the purpose).

Well, don't be so quick.  The problem is that I use global counter
that gets incremented after EVERY query.  So e.g. when you have this
data:

@ MX 10 a.example.com
@ MX 20 b.example.com
@ A 127.0.0.1
@ A 127.0.0.2

and are asking for ANY RR, you'll get this:

example.com.            2100    IN      A       127.0.0.1
example.com.            2100    IN      A       127.0.0.2
example.com.            2100    IN      MX      20 b.example.com.
example.com.            2100    IN      MX      10 a.example.com.

This is because the counter gets incremented twice, and as a result
rbldnsd will return the same result set a1ll the time.  That's why
i called the "randomization" as "dumb".  But at least, this is a
start... ;)

/mjt

P.S.  After mailman upgrade, From: field in this mailing list looks
somewhat differently... ;)  Anyone know how to "fix" it back, to stop
mailman adding the comment?