MAX_RPL_WHO is hard-coded

[alexbarton]

(Report imported from Bugzilla #125)

Status CLOSED, severity normal, in component Daemon.
Reported in version 19.1 on platform All.
Assigned to: Alexander Barton.

On 2012-08-16 15:11:12 +0200, Michael Stowe wrote:

A lot of irc clients use /who list commands to figure out who is on the server, routinely. The hard-limit in MAX_RPL_WHO means that some clients get spammed with "NOTICE ... list limit (25) reached!" which some users find obnoxious.

I'm not sure why it's limited at all, but having a configurable limit would ease the pain. Or maybe remove the notice for /who commands, I'm not sure how other servers handle it, but this notice is unique to ngircd (and intensely disliked, apparently.)

On 2012-08-18 23:45:30 +0200, Alexander Barton wrote:

The limit has been introduced by commit c5beca8 to prevent clients from flooding the server.

So now there are a few options:

1. ignore these clients – not a good idea ;-)
2. remove this limit all together – most probably a bad idea, too …
3. set the limit higher – but to which value?
4. make it configurable …

Any thoughts?
Most probably WHO is called for a channel ("WHO #channel") in this case, right?

On 2012-08-19 00:08:31 +0200, Michael Stowe wrote:

Yes, it looks like a WHO #channel command -- I assume some clients issue one continuously to keep their channel lists updated (or maybe it keeps getting re-issued because they don't like the error.)

I'd go for option 3 or 4 -- almost arbitrarily, I'd pick 200, since a lot of channels have more than 25 people in them but very few have more than 200.

Configurable would work well, too, but clearly it would be more work to implement.

On 2012-09-10 12:48:11 +0200, Alexander Barton wrote:

I thought about it a bit, but in the end, I think "WHO #channel" makes no sense at all if it wouldn't report the complete list. So I removed the limit. But if there are more than MAX_RPL_WHO(25) replies, the client requesting the list will be "penalized" one second more (then 2 in total).

Commit 2205227.

On 2012-09-10 13:11:06 +0200, Michael Stowe wrote:

That seems eminently sensible to me

On 2012-12-31 00:20:46 +0100, James Kirwill wrote:

*** Bug 148 has been marked as a duplicate of this bug. ***

On 2013-02-11 22:33:56 +0100, Alexander Barton wrote:

2205227Patch included in ngIRCd 20.
Bug closed.