On Wed, Dec 19, 2012 at 02:20:23PM -0600, lists@packetmail.net wrote:
On 12/19/2012 02:17 PM, Alexander Barton wrote:
The same password and exactly this command sequence works (copy-paste!) for me, so ... WTF!?
Haha, crazy!
Yeah. Me. Sorry for the confusion, but I was testing an old binary ... waaaaaaah :-(
I _can_ reproduce your problem.
And yes, ngIRCd 20 behaves differently: it tries to compare the cut off server password with the original "long" client password, which then isn't equal. This has been changed in commit 7df4c12d, "Dynamically allocate memory for connection password", and is the "Rigel thing" to do ...
So basically there is no other option than to provide the correct password that the daemon uses -- that is, the cut off password.
The only alternative would be to allocate the server password dynamically, too, and therefore to allow server passwords of arbitrary length ...
Alex