What about +d?

General talk about EFnet

Moderators: Website/Forum Admins, EFnet/General Moderators

Should we enable +d?

Yes
20
69%
No
6
21%
Not bothered
3
10%
 
Total votes: 29
User avatar
HM2K
Posts: 209
Joined: Thu Jul 24, 2003 5:34 pm
Location: UK
Contact:

What about +d?

Postby HM2K » Wed Aug 20, 2003 10:46 am

I can honestly say that I think +d would be more useful than +e and +I put together, and I'm sure you could figure out why...

For those who don't know, +d bans the gecos/name in a channel, ie: "/mode #efnet.org +d gettiniton"

I know I can't get an official answer here, but I was wondering the kind of responce I would recieve on this...
User avatar
Jon
Posts: 42
Joined: Wed Jul 16, 2003 5:33 am
Location: NB.CA

Postby Jon » Thu Aug 21, 2003 4:23 pm

I recall someone talking about +d would be better, not sure if it was in this forum or another (too lazy to check)

I actually forgot what +d does, and your definition wasn`t very clear... I`m not certain what a "name" is. Is that the "real name" that you see in a /whois ?
User avatar
lucy
Posts: 234
Joined: Wed Jul 02, 2003 6:22 pm
Location: graceland
Contact:

Postby lucy » Thu Aug 21, 2003 5:16 pm

Jon wrote:I actually forgot what +d does, and your definition wasn`t very clear... I`m not certain what a "name" is. Is that the "real name" that you see in a /whois ?
umm i thought it was just for the real name, or full name as its called in mIRC, but I just found this on the irchelp.org website.... so maybe its for more than just the real name field.

"New channelmode +d
This is an extended form of channel bans. These work on wildcarded strings, not hostmask. They will match anything in the users /whois , including servername, nickname, username, hostname, and realname. Example: /mode #channelname +d *John*Doe* . Prevents users with "John" and "Doe" anywhere in a whois field from joining #channelname. This is useful against certain types of clone attacks, as well as useful for denying access to a channel via a certian IRC server. Use this with care. Also not that it will not work completely unless all servers on the net are Hybrid 6."
leeh
ircd-ratbox coder
Posts: 48
Joined: Wed Jul 02, 2003 5:43 pm
Location: UK

Postby leeh » Thu Aug 21, 2003 7:02 pm

lucy wrote: umm i thought it was just for the real name, or full name as its called in mIRC, but I just found this on the irchelp.org website.... so maybe its for more than just the real name field.
The original hyb6 implementation (which was later removed) worked just on a clients gecos field. It is easy to extend to ban clients on a certain server as well however.

Personally, I dont think its worth it. Theres short term gain granted, in the long term you dont achieve much. A lot of drones have already adapted to bans on gecos fields done by servers - id expect them to do the same should channels get +d.

They might be useful if you had regexp support, but I doubt enough people would be able to use regexps to make it worthwhile. :)
User avatar
lucy
Posts: 234
Joined: Wed Jul 02, 2003 6:22 pm
Location: graceland
Contact:

Postby lucy » Thu Aug 21, 2003 7:18 pm

leeh wrote:They might be useful if you had regexp support, but I doubt enough people would be able to use regexps to make it worthwhile. :)

you should know someones gonna ask what regexp support is.... so go ahead and tell us and dont make me look it up :)
User avatar
munky
Site Admin
Posts: 826
Joined: Wed Jul 02, 2003 4:54 pm
Location: Phoenix AZ
Contact:

Postby munky » Thu Aug 21, 2003 7:56 pm

regular expressions

^P[a-z]{2}[0-9]{3,4}$
would match strings that start with P, have 2 lower case letters, and end with 3-4 numbers
In God we trust,
Everyone else must have an X.509 certificate.
Hwy
Posts: 66
Joined: Wed Jul 16, 2003 12:27 pm

Postby Hwy » Thu Aug 21, 2003 8:44 pm

But also remember the CPU load regular expressions would impart on a server. The original +d in Hybrid 6 used regular expressions (early betas) but it was changed to match() (wildcards * and ?).

What WOULD help and not cost much is an improved match(), like oomon's current implementation. You would have additional matching characters, such as # for a digit.
User avatar
Gozem
Posts: 22
Joined: Mon Jul 14, 2003 6:42 pm

Postby Gozem » Fri Aug 22, 2003 2:12 am

Hwy wrote:But also remember the CPU load regular expressions would impart on a server. The original +d in Hybrid 6 used regular expressions (early betas) but it was changed to match() (wildcards * and ?).

What WOULD help and not cost much is an improved match(), like oomon's current implementation. You would have additional matching characters, such as # for a digit.
Saving CPU in all honor but really, the server today on EFnet uses sub 2-3% of the CPU in average.
oper, irc.csbnet.se
User avatar
strikelight
Posts: 31
Joined: Tue Jul 08, 2003 1:13 am
Location: Canada
Contact:

Postby strikelight » Fri Aug 22, 2003 4:04 am

Man, I've been calling for the re-introduction of +d since it first was retired from EFNet after being 'unofficially' introduced.... Save your breath, if it hasn't happened since then, they obviously care not to bring it back. :(

Code: Select all

 ____________
|:   _____  :|
|___    |   .|____
|:  |  :|    |   :|
|_______|_________|
[/color]
- silence isn't golden when i'm holding it inside -
Hardy
Site Admin
Posts: 394
Joined: Wed Jul 02, 2003 4:54 pm
Location: Oslo, Norway
Contact:

Postby Hardy » Fri Aug 22, 2003 11:43 am

strikelight wrote:Man, I've been calling for the re-introduction of +d since it first was retired from EFNet after being 'unofficially' introduced.... Save your breath, if it hasn't happened since then, they obviously care not to bring it back. :(
Adding new features on EFnet is quite a job. First of all you need to get the coders of several ircd`s to code it (csircd, ratbox & hybrid ), and then also create patches for older versions of these ( hybrid7, hybrid6 ). Then every admin need to implement this and restart their ircd...

Efnet doesnt have one uniformed ircd, and doesnt require admins to upate to newest version of something.. if a admin wanted to run hybrid5 I doubt anyone could object if it had patched in TS5 support..

Changes are good, but we cant add a new mode every month. This time admins agreed on numbers of bans, +I and +e.. perhaps +d is inserted in the next vote which i hope it takes a looong time until is casted :)
-- Hardy
Administrator: irc.underworld.no
Services Administrator
http://www.efnet.org admin/staff
User avatar
HM2K
Posts: 209
Joined: Thu Jul 24, 2003 5:34 pm
Location: UK
Contact:

Postby HM2K » Fri Aug 22, 2003 11:56 am

+I and +e seemed to go through fairly quickly, as soon as I hear about it, they are being implimented... I can't see that its that much of a problem to impliment +d, just more of an excuse not to... You say that coders would have to change their code, but as far as I know, it is/should already be implimented, it simply needs to be enabled in the config, but I don't know, i'm no expert.

I see why +I and +e have been introduced, since they are pretty much a standard these days, with even mIRC recognising that in its channel properties... but I feel that +d would come in handy, maybe only in the short term, but who really knows the long term effects? but to be honest, I really can't see +d causing any problems, it may even be part of the solution to a few things.
User avatar
munky
Site Admin
Posts: 826
Joined: Wed Jul 02, 2003 4:54 pm
Location: Phoenix AZ
Contact:

Postby munky » Fri Aug 22, 2003 12:34 pm

Gozem wrote: Saving CPU in all honor but really, the server today on EFnet uses sub 2-3% of the CPU in average.
true enough, but depending on the implementation of regexp, it can be a huge hit. can you imagine what would happen on a server with 10000 users if every connect/join was passed through a perl mod? (don't get me wrong, perl regexps are great, but the script interpreter isn't always the quickest) especially if someone did crazy stuff or wasn't good at writing efficient regexps.
In God we trust,
Everyone else must have an X.509 certificate.
User avatar
Gozem
Posts: 22
Joined: Mon Jul 14, 2003 6:42 pm

Postby Gozem » Fri Aug 22, 2003 5:44 pm

munky wrote:
Gozem wrote: Saving CPU in all honor but really, the server today on EFnet uses sub 2-3% of the CPU in average.
true enough, but depending on the implementation of regexp, it can be a huge hit. can you imagine what would happen on a server with 10000 users if every connect/join was passed through a perl mod? (don't get me wrong, perl regexps are great, but the script interpreter isn't always the quickest) especially if someone did crazy stuff or wasn't good at writing efficient regexps.
Of couse is the regexp-interpreter written in C-code.
oper, irc.csbnet.se
leeh
ircd-ratbox coder
Posts: 48
Joined: Wed Jul 02, 2003 5:43 pm
Location: UK

Postby leeh » Fri Aug 22, 2003 7:32 pm

HM2K wrote:+I and +e seemed to go through fairly quickly, as soon as I hear about it, they are being implimented... I can't see that its that much of a problem to impliment +d, just more of an excuse not to... You say that coders would have to change their code, but as far as I know, it is/should already be implimented, it simply needs to be enabled in the config, but I don't know, i'm no expert.
It was actually a couple of years for +e to get through efnet - itd been around since hyb6.0 which was eons ago. :)
prefect
Posts: 76
Joined: Mon Jul 14, 2003 6:25 pm
Location: Oslo

Postby prefect » Sun Aug 24, 2003 12:40 pm

I don't remember if +d had this functionality, but it would be very useful to have a channel mode to block out users on specific irc-servers. We've all experienced trojan/drone/proxy-kiddies taking advantage of servers with no active opers (no active opers there and then), and it would be cute to have the possibility to ban the whole server temporarly.

Or ircadmins could be better at adding eachother as tcmopers or whatever, not having active opers on a server that allows clients without ident and/or hostname is asking for trouble.

Edit: edited because it's early in the morning.

Who is online

Users browsing this forum: No registered users and 2 guests