Closed Heufneutje closed 10 years ago
Whoops. Didn't mean to hit that.
UUIDs are the general fallback nick, and we use it in a lot of places (resolving nick collisions, for example). UUIDs are internal use except when the server needs a backup nick to give to someone (in cases like here or nick collision on netmerge, for example). Having a default prefix (nothing) as something that cannot collide with anything saves us a potential step there, since if we hit a collision with the prefix we still fall back onto the UUID. We don't go as far as InspIRCd and expose NICK 0, but using the UUID is still useful.
Also: I'm glad the other one ended up closed, because we weren't going to go back to random number generation.
Regardless, the bugfix this PR adds should be implemented.
Also, I don't see how adding a prefix by default is a bad thing. If the UUID is appended it's always unique anyway.
Also, having a non-empty default is more similar to the intended use case for DB, ie. Busser.
If the UUID is appended it's always unique anyway.
No, with a prefix on the UUID, it's a valid nick that someone else could've changed to. (It's unlikely, but possible.)
Also, having a non-empty default is more similar to the intended use case for DB
DB is probably going to change a fair amount of configuration (especially relating to services).
No, with a prefix on the UUID, it's a valid nick that someone else could've changed to.
Yeah, but even if that happens it falls back on the UUID anyway... I really don't think having a default of "Guest" is a bad thing.
But still, it's preferable if we can stick to a nick like Guest0AAAAAAAA instead of just 0AAAAAAAA. The overhead of the extra check is 100% negligible. On 5 Sep 2014 16:31, "Jacob Riddle" notifications@github.com wrote:
If the UUID is appended it's always unique anyway. No, with a prefix on the UUID, it's a valid nick that someone else could've changed to.
— Reply to this email directly or view it on GitHub https://github.com/ElementalAlchemist/txircd/pull/53#issuecomment-54693787 .
Follow-up on #52 as GitHub wouldn't let me reopen the PR after I ammended my commits. This fixes the bug I mentioned there as well. I'm still iffy on the UUIDs as I think they should be a thing that's, at least for the most part, internal use only. I don't have a better solution as of right now though.