Here is my blurb of the day. It is about communication media in Opensource development.
Yesterday we talked on #kde-freebsd about the k-f teams release process. The consensus was, that we need to get more people into the team. Lofi has posted a job advertisement the Mailinglist. He said that feedback has been very low.
I pointed out, that in my opinion too much communication is done via IRC, and most people (including myself) don’t have the time to be on IRC the whole day. If there was actual more ports-preparing related traffic on the Mailinglist, more people would get involved.
I was opposed by nearly everyone, the main point was, that if all communication was done via the mailinglist, we would still be working on KDE 3.2, since communication via mailinglist is extremly slow and inefficient.
So Question to the reader. If you are interested in an Opensource Project. Would you:
a) Join the IRC channel and talk to the developers directly
b) start lurking on a mailinglist.
I can only tell from my own expierience. I maintain around 30 ports. It the project has a Mailinglist, I usually join. If it has a developers list, I browse the archive for interesting threads. I have one port that has only an IRC channel for getting help, talking to the developers and stuff. I have had a lot of problems, getting the attention of the developers. On third of the times I joined the channel they weren’t there or sleeping. Another third, they are debating someone else’s bug or other important development things, and the other third they are in an offtopic debate, or drunk, or tired or whatelse. Another problem is to determine if someone on IRC is one of the developers or just a troll. On Mailinglists, people can be identified by Email addresss (FreeBSD.org -> FreeBSD developer, KDE.org -> KDE member) or they have signatures telling everybody what they work on.
I admit I am an IRC addict, I am usually in at least three channels, and nearly all the time I am sitting at a computer I have an IRC window somewhere. It is the fastest way of communication with someone on the other side of the globe.
It is a nice environment for getting someone to know. BUT I usually hang out in my favorite chans, because I know whom I talk to, and the signal2noise rate is okay. I join a different channel only if someone I know is on there, but not because someone unknown invites me to join.
IMHO the main disadvantages of IRC in regard to serious communication/development are the real-time (“XY is currently sleeping, are we waiting for his review, or are we just going ahead?”), the missing archives (Who wants to ready IRClogs from 10 years ago? Who wants his private IRC talk somewhere on the web?) and the noise rate (There are so many idiots on IRC, usually every channel has at least one).
People say mailinglists are slow. Yeah, usually people check email only every 10 to 30 minutes, so you can’t expect immediate feedback. But there are major advantages, like threads and subjects which enable you to talk about two different things at the same time without getting confused. Email-addresses, which offer the possibility to ask questions after years. And with Attachments you can attach patches directly to the communication.
Is the slowness of mailinglists really a killer argument? IMHO no, we have all the time in the world, since we don’t need to sell our software. So I think that Mailinglists are superior.
What is the best way to combine IRC and Mailinglists in Opensource Development? I think it is a good idea that it should be possible for people who are not on IRC to follow the development. This means: Everything that should be reviewed, like patches, design documents, TODO lists etc. should be posted to the mailinglist. Discussions could happen on IRC, but decisions should be made on the mailinglist.
My favorite example: FreeBSD. FreeBSD has no official IRC channel. But of course you can meet a lot of developers on channels like #bsdcode. FreeBSD newcomers usually start following the mailinglist, and writing bug reports. If they are interested they usually start talking to developers on IRC.
Please comment if you think, this makes sense, or if you think i am on the wrong way.
16 thoughts on “IRC vs. Mailinglist”
Why don’t you post this to a mailing list? 🙂
I agree. I even tried it, before this last release, posting more to the k-f-devel list, and avoiding the IRC channel.
The real-time aspect of IRC, is a tricky thing. It’s as much a blocker for getting things done, as it is the addictive part of why we end up communicating that way.
ever thought about a wiki to support you?
Ugh. The Wicked Wiki Word.
We could have a joint project blog though 🙂
ad le. Valid point. I was writing this article, to get my ideas straight.
I think I will not post it to the Mailinglist, Since I discussed this on IRC, and most people that count had a different opinion.
If everyone else thinks, that things are well, why should I start
spreading bad mood by starting a bikeshed-thread.
ad lauri. I think the k-f-devel list, pretty much shows whats wrong with the mailinglists. It was created to seperate the devel-traffic from the general help & updating traffic. But so far, 90% of the traffic are the automatic posts from the CVS builds.
I think a wiki isn’t really what we want. Since all members have CVS accounts, we could just edit the normal website. We have already something like that, the dev-docs module
Wiki is ideal if there are many anonymous contributors. But here we have a nearly closed usergroup.
ad lauri 2.
I know this was more fun, but I kind of like that blog idea. From time to time I browse the DragonFlyBSD blog and the kdedevelopers-blogs. We just have to persuade the anti-blog fraction (ade & tap) of the k-f team 🙂
I like IRC as a means of communication, but a groupware website or such would probably be apropriate as well for this purpose.
I hate mailing lists myself, and only use http frontends to snag important threads if referred there from IRC, or if they are referenced in a human-created digest.
Wonder how long it would take to write some custom CMS software that does ML integration.
This is interesting. Why do you hate Mailinglists?
And do you know any serious Opensource-project, that uses a “groupware website” for communication with each other? I am not talking about this phpbb-Boards. They are just ugly.
Since I somehow started and inspired this whole thing, a late comment for me: For me, it’s really not about IRC vs. Mailinglists – the mode of communication in a project is just a reflection of the people involved, and the number of people involved.
A project consisting of sufficiently few people, which are all IRC regulars anyway, will naturally have a communications focus on IRC. There is no way a project will _stay_ focused on IRC if mailinglists are available (even if at first pretty much unused) AND:
– The number of people involved grows beyond a certain limit
– The people involved are beginning to spread across timezones so that important contributors will almost always miss each other on IRC
– More people who don’t do IRC anyway (those exist) contribute to the project
I’m really not favouring any particular mode of project communication, I just go along with what works. Times might come when I won’t be able to spend much time on IRC anyway (jobs, etc), and while I will certainly not cease project work, I’ll become a heavy-ml user. 🙂
Again: I think it’s really not the comms-hub that influences the project and the people involved, it’s mostly the other way around.
Oh yeah, and that groupware idea sucks, I’m really really opposed to _that_. Just because it’s a common means of work^W torture in corporate communications doesn’t mean it’s any good.
“the mode of communication in a project is just a reflection of the people involved”
Yeah, thats probably right. But, is the mode of communication influencing how many and what kind of people join an opensource project? Are we attracting potential developers by pointing them to our IRC channel?
I admit, I have lost the point in the second part of my article.
The blog suggestion was only partly in fun. I think it could be at least as useful as a wiki, and a heck of a lot friendlier to work in (especially if you take a look at the ‘Kajsa’s corner’ website of mine, which is to all intents and purposes turning MT into a straight up CMS. Especially look at a page I’ve put actual content on, like, Trollmorin the lullabies menu.) It’s going to be pretty obvious to anyone familiar with MT what’s behind the site, but a little CSS would hide that (I’ve just been too lazy to so far.) It’s probably not obvious otherwise, that it’s not your average plain website, at least until you hit a page that’s allowing comments.
In any case, I think “blog” becomes overloaded for many people, the way “wiki” is for me – I know for certain there are useful, well used, user friendly and easy to customize wiki sites out there. I can’t get past the cluttered, awkward, nasty versions I see most people set up though, with way too much stuff on.
“Pages” are posts to a topic, a bit of wrangling with the templates and you get a categorised menu (and, niftily, it simply skips entries that aren’t sent to one of the four specific categories that are set to be menu entries, so you can just as well use it to do the usual ‘diary’ type stuff, without it showing up down the side. This is going so far off topic as to be ridiculous 🙂
Heh, I think I’ve tried nearly every groupware, trouble tracker, whatever software there is out there, and the nearest we’ve come to anything useful, is the devdocs.
Maybe I can set up a multi-author blog-come-CMS with integrated developer and/or development diary that also integrates the content from the devdocs site – thus pleasing the half of us who can manage editing plain text in CVS but hate blogs, as well as the half of us who actually do like blogs, as well as the other half who just avoid anything that might cause them to have to do work 🙂
(yes I know, I got that up to three halves, you all know I don’t do math)
Lauri: how about this — http://tblog.ath.cx/wiki
That site isn’t about anything, more of just a ‘throw it up there and see what happens’ that is powered by exactly the same software that powers my blog.
Due to the modularity, I could easily write ML or CVS interfaces to it without complicating the existing work. It also has ACL’s and other toys that I’ve programmed into it so that it might be useful someday for something more than just a blog. (works pretty good as a forum and wiki already)
I agree – not sure anymore with whom and what though.
Blog software is not much different from a CMS, and comments are not much different from a forum.. etc etc. My blog/CMS system pretty much has what lauri describes: a web form to post new entries, a cronjob reading external RSS feeds to import new entries and an e-mail interface so I can even post new entries by mailing them (and even though there is some authentication in there, let’s not make that address public heh). All supporting multiple authors of course.
He, looks like the real 1337 people have written their own CMS. I feel pretty lame with using MT 🙂
1337 indeed — I wrote it though because I was fed up with kword and wanted a word processor that dealt with hyperlinked reality rather than flat pages. Really, it just accidentally evolved from there into a blog/CMS.
Comments are closed.