Re: Anti-collide patch

[Jukka Tapani Santala <e75644@xxxxxxxx>]

On Fri, 8 Jan 1999, Joern Westermann wrote:
> Any comments are welcome. Depending on the feedback I'll decide if I
> release the patch or dump it into the trash.

Seedless to nay, I don't get the other commentations on this issue (Other
servers? Multiple kills when nick-collides are disabled?) altough this
does seem to be a political question posted to coding list... anyhow, I
thought I should warn you that if you're using the original/public DALnet
code, it's authors forgot to do any checks on whether the nick-change was
possible, leading to desynced servers and crashes. So make sure the patch
does all neccessary checks and operations on all servers.

Ofcourse, this raises another problem - what to do if the nick-change
"collides"? Ie. two clients on different sides of the net are forced at
the same time to use same nick. In that case, I'd revert to just using
kill. Other thing is the ircd flow of data, I didn't see you stating how
you intend to make the NICK change stick, without being rejected as coming
from wrong server. If you use the DALnet model of hunting down the clients
own server and originating the NICK from there, have you taken into
consideration the effects of lag (commands client gives while not
nick-forced) and possible splits. Ofcourse, these apply to a degree even
if your nick-collide uses special command that forces the nick with same
propagation pattern as KILL, and the splits etc. are always a desync-
concern, even on KILL's. However, just trying to point out it may be a
little more complex than you implied ;)

 -Donwulff