Tree [8e6c65] alpha_0 /

HTTPS access

File Date Author Commit
 config 2001-03-08 Joseph Bongaarts Joseph Bongaarts [a42e20]
 doc 2001-04-15 Perry Lorier Perry Lorier [241fa0]
 include 2001-04-15 Perry Lorier Perry Lorier [241fa0]
 ircd 2001-04-15 Perry Lorier Perry Lorier [241fa0]
 tools 2000-04-13 Greg Sikorski Greg Sikorski [a5a606]
 .cvsignore 1999-11-16 Bleep Bleep [b70944] This commit was generated by cvs2svn to compens... 1999-11-16 Bleep Bleep [b70944] This commit was generated by cvs2svn to compens...
 BUGS 2000-07-15 Perry Lorier Perry Lorier [5713fa]
 ChangeLog 2001-03-14 Joseph Bongaarts Joseph Bongaarts [aa5638]
 ChangeLog.07 2000-03-18 Bleep Bleep [ae91ef]
 ChangeLog.10 2000-09-17 Bleep Bleep [f0e781]
 INSTALL 2000-03-18 Bleep Bleep [ae91ef]
 INSTALL_FR 2001-03-31 Perry Lorier Perry Lorier [8af0b3]
 LICENSE 1999-11-16 Bleep Bleep [b70944] This commit was generated by cvs2svn to compens... 2000-12-16 Perry Lorier Perry Lorier [30cecb]
 README 2000-03-18 Bleep Bleep [ae91ef]
 RELEASE.NOTES 2001-04-15 Perry Lorier Perry Lorier [241fa0]
 TODO 2000-07-15 Perry Lorier Perry Lorier [033ec6]
 configure 2000-04-07 Kevin L. Mitchell Kevin L. Mitchell [e8dbd1]

Read Me

README for ircu2.10.10.beta
Please read this completely before running the server

* If you run ircu2.10.10 on the production network and expect
  more than 1000 local clients to connect you will want to make
  sure you add -DNDEBUG to the extra CFLAGS setting when you do
  a make config.
* Ircu now uses asserts in a lot of places to insure referential
  integrity and that impossible situations do not occur. Also
  memory integrity checking is defaulted to on as well. The asserts
  and memory checking both use cpu and in the case of memory checking
  extra memory. We have not had an assert trigger in quite some time
  but would prefer to leave the checks enabled for beta testing.
  The asserts insure that if the server is going to core it will do
  so in well defined places where the problem will be easy to trace,
  without them the server could core in not so well defined places.
  If you want to disable the checks, add -DNDEBUG in the extra CFLAGS 
  when you do a make config.

* Ircu no longer uses separate C/N lines, the functionality of both
  has been combined in C:lines, see the example.conf for more information.

* Ircu has an different mechanism for defining ports than the previous
  versions. Please see the example.conf for information on configuring
  them. Ircu also ignores the port specified on the M:line, you MUST define
  a server port with a P:line if you want servers to connect to you.

* For beta releases, please understand that there may be bugs we haven't
  found yet, (that's why you're beta testing it in the first place :)).
  Please report any bugs found in the server to or with as much information as you can provide
  about how to reproduce the problem. Stack traces for coredumps are
  usually helpful, however we aren't expecting any of those ;-).

* Ircu 2.10.08 is a P10 only server, and cannot be used to host P9 
  links. On a mostly P10 network this is not a major issue, it does
  mean that you cannot connect services directly to a 2.10.08 server,
  but does not preclude it's use as a leaf or a hub. For now, services
  need to continue to be hosted by 2.10.07 servers.

* EXTENDED_NUMERICS is not defined by default, we need to get everyone
  using 2.10.07 before we can allow it to be turned on anywhere. If you
  are using 2.10.08 on a network where everyone is up to at least 2.10.07
  it is safe to turn them on. We have tested them on testnet and they do
  work. :)

* The masked notices using #*.mask has changed to use $@*.mask instead.
  Since most of the servers on the net will not understand the new masking
  mechanism, this probably won't work for a while. The server notice mask
  ($* still functions properly. This was a trade off
  between backwards compatibility and the improved protocol, we chose to
  not maintain backwards compatibility for this command. The #*.mask
  functionality is easily circumvented, and the command is rarely used
  it is not considered to be a major issue.
* The server no longer sends notices for errors on connections that are
  suddenly dropped during connection setup. If you really want to see
  them, we suggest you get a life. :)

* Removed: unix domain sockets, uping, m4 preprocessor spawning,
  dozens of bugs.

* New stuff:
   Added ISUPPORT code to match dalnets.
   Added connection progress notices.
   Cleaned up operating system checks.
   /invite with no arguments lists the channels you're invited to.
   Many speedups, ipdiffer, faster strncpy, faster inetntoa.
   Passive lag and numerics now displayed by map command.
   Server to server tokenization.
   Much larger TCP windows/kernel buffers for server connections,
    smaller for clients.
   Removed BOOT_OPER security hole.
   Complete listener port specifications including per port virtual
    hosting and hidden listeners.
   Default server port for connects (config option).
   Speed up some channel ops using new Membership struct.
   Hash table performance stats are available now (basic).
   Much of the socket code has been rewritten, we now only make one
    pass through the local client array.
   Added initial support for PROTOCOL command.
   Many functions have been made reentrant.
   Resolver now uses callbacks for all queries.
   Almost every command that is sent to a server uses full P10 numerics
    and tokenization, this reduces netburst sizes by at least 10% and
    many server to server message sizes by 10-50%.
   Major source code reorginization to support engine model.
   Parser now uses an indexed table for commands, which elminates
    the need to do IsServer IsUser IsOper checks and allows much more
    efficient implementation of user/server/oper command handlers.

* For a much longer winded explanation of all this stuff, see the
  ChangeLog in this directory, or the cvs log at the coder-com
  web site at

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.

No, thanks