IRC-Junkie.org – IRC News

All about Internet Relay Chat

freenodes ircd-seven is in the final testing stage

ircd-seven, the IRCd that is going to replace the aged hyperion IRCd currently in use on freenode, is in the final stages of testing.

After being in the public testing phase for over one year, it seems it is finally ready to go into production use on the network.

In an announcement, freenode staff write that after “extensive testing by users and staff, we are now preparing for the switch-over which is taking place at the end of this month” and that they’d like to thank “those of you who have helped test, those who have botted the testnet and in particular those who have helped us find and iron out bugs”.

Those who make use of bots on freenode are advised to “take these last couple of weeks to make sure that they work with the new ircd, so as to not experience disappointment on switch-over for the production network”.

Closing the announcement they write that if “all going well, the switch-over is scheduled for Saturday January 30th 2010″.

Category: IRC, IRCd, Networks, Software
  • katsklaw says:

    Why can’t developers just leave things alone? Just because something is “aged” does mean it’s inferior. Some of the best, most stable and proven irc software on the planet is older than FreeNode it’s self and still handling loads larger than FreeNode.

    I’m sure devs are thankful their employers don’t use the same practice on their careers “I’m sorry Mr Dev, you’ve been here for 2yrs, I’m afraid you’re “aged” and must be replaced.”

    If it’s broken, fix it. If not, leave it alone!

    January 13, 2010 at 2:32 pm
  • nenolod says:

    Actually, katsklaw, development of freenode’s next-generation ircd has been in progress for half a decade. ircd-seven is the freenode-specific parts of charybdis that haven’t been merged for whatever reason. Those parts will eventually reach near-zero, and basically have for the most part.

    The ircd software you speak of, be it ircu or ratbox, or whatever has benefited from our work on charybdis, and it’s branches (sorircd2, seven, etc) in terms of mass stability and robustness increases. If you’re talking true 2.8-based IRCds, well, you are right. Hyperion is indeed more reliable than an IRCd closer to the 2.8 tree. However, the reason for that is mostly due to the fact that IRCd coders are rarely serious about producing a product that actually works.

    For example, some of the features you enjoy in ratbox3 on efnet were incubated in charybdis.

    Anyway, back to your “If it’s broken, fix it. If not, leave it alone!” thing. Hyperion was broken from day 1. The codebase was so awful that not much could be done to fix it, so we fixed the major problems and started working on what became charybdis. What became charybdis was originally a freenode project, but ultimately jilles and I decided the work was more suited under the atheme umbrella than under freenode/PDPC’s at the time. A large part of that had to do with IRC politics which made collaboration with lilo no longer a feasible thing (if you want to know more about that, you know where to find us). After lilo died, freenode tried to collaborate with OFTC for a while, but quickly found out that collaboration with atheme was the best way to go. Blah blah blah, my point is, hyperion was broken from day 1. It was meant to be a stop-gap. Unfortunately freenode has used it this long because when we were working on initial development of charybdis, lilo was uncooperative in participating in the process, which caused us to restructure the project. In the last 3 years, we have done great work collaborating with freenode, such as extending, augmenting and improving the atheme services package for maximum scalability. In the last 2 years, we have collaborated on seven, and it has been extremely effective. So basically, charybdis/seven is the fix for hyperion.

    January 17, 2010 at 7:27 am
  • katsklaw says:

    Actually i was referring to Bahamut. Which has been proven stable thus development slowed considerably. Yes, it replaced DreamForge, but not because DF was old, it was because it was inferior and couldn’t handle the ever increasing needs of DALnet. However, ircu, is a good example too :)

    As far as the horrible code base, reminds me of another popular ircd/services package or 3. No one likes digging around in an undocumented, convoluted API or core so I’m with you there. Anyway, my last comment did come out harsher than intended and I do apologize for that. However, this *is* just one instance. It seems that “rewrites from scratch in ” seems all the rage anymore. Which is what triggered my reply in the first place :)

    February 6, 2010 at 3:08 am

Your email address will not be published. Required fields are marked *

*