FreeBSD Core Developer Thrown Out

SlashChick writes "From a discussion on the freebsd-chat mailing list, it appears that one of the FreeBSD core developers, Matt Dillon, has been barred from committing any changes to the FreeBSD kernel. Dillon was one of the developers 'responsible for making FreeBSD 4.x the most rugged and stress-proof free operating system in existence,' and also contributed to fixing the Linux VM. Unfortunately, there has been little explanation from the FreeBSD core team about why Dillon was thrown out, leading to speculation and worries about the future of the FreeBSD kernel. Does the Slashdot community have any more insight into this situation? Would someone from the FreeBSD team care to elaborate and assuage our worries?" CD Update: Greg Lehey from the core team has infact elaborated in this comment.

Little explanation? I think there's enough.

[angio]

From the mail archive:

The short of it is that Matt was unable to treat many of his fellow developers with the civility and respect that they deserve.

I think that's fairly clear. There are many strong, good hackers in this world who wouldn't be able to work together. While it's unfortunate that Matt and the rest of -core weren't able to resolve it, it's a fact of life in a big project...

Re:Nice name...

[Anonymous Coward]

Matt Dillon did his C compiler for the amiga, when the actor was a teenager star...

Re:TdR

[Anonymous Coward]

Theo's version of events: http://www.theos.com/deraadt/coremail.html [theos.com]

Yes, THAT Matt.

[Anonymous Coward]

Did you happen to use Matt's Amiga development tools? DCC was an excellent compiler--and cheap, too.

Matt is no slouch. Glad someone else remembers him--a refreshing change of pace from the zero-history PC heads on Slashdot.

Re:Little explanation? I think there's enough.

[RevAaron]

Just as 5.0 was coming out and there seemed to be this general quietness about freedom and the GPL and RMS bashing seemed to be at an acceptable high point the good lads at FreeBSD go and remind us all again what open and freedom is all about.

What does this have to do with GPL or its attitude? Linus could just as easily bar kernel patch submission from some individual who he thought was causing problems.

This problem individual could just as easily keep on running with his own special kernel, with all these swell changes Linux kept rejecting. Same goes for FreeBSD- you can get the source, and this guy, or anyone, could keep on writing new code, patching their setup and giving the away the code.

Just because a project is GPL doesn't mean that it'll take code from anyone, or have a CVS server to which anyone could commit. From where would you get this silly idea?

dillon leaves the FreeBSD project

[groggy-P]

The information you have is from the open FreeBSD-chat mailing list. The thread was started by people in the project who are not on the closed developers@FreeBSD-org mailing list, and some of the statements are wide of the mark:

• Matt Dillon was never a "core developer". The FreeBSD project doesn't use that term, but it looks like a reference to the core team [freebsd.org]. Matt has never been a member of the core team.
• Matt has done some very good work over the years. His contribution to FreeBSD release 4 was invaluable, but it would be wrong to suggest that he single-handedly made the difference. Commit statistics on the orginal list show that he has not been very active over the last 12 months.
• I was not aware of his involvement with Linux VM. Nothing we have done will change this, though.
• The FreeBSD core team has informed the development community in detail about the reasons for Matt's removal. We don't think it's appropriate, nor fair to Matt, to wash dirty linen in public.
• Matt has very little influence on the future of the FreeBSD kernel. That work which he has done over the last two years or so was mainly maintenance.

It's always sad to have to make these decisions. It's even more difficult to defend them when our hands are tied behind our backs.

Re:Bona fide cuestions

[mph]

Who elects core developers?

The committers (i.e. developers with write access to the CVS repository, who number in the hundreds).

If core developers are elected democratically, why was it necessary to throw out Matt Dillon explicitly instead of just explaining the situation to the electors? Aren't electors trusted to do the (apparently) reasonable thing and unelect him?

Contrary to the title of this Slashdot article, Matt Dillon was not a member of Core, and hence was not elected. He was a committer, one of hundreds. His commit bit was granted by Core.

It is incredibly frustrating to read Slashdot whenever something like this comes up, because so many people (Linux people, it seems) confuse Core with the body of committers, despite the fact that we go through this exercise over and over again.

Dillon Interview

[Anonymous Coward]

For those unfamiliar with the afformentioned Matt Dillon, KernelTrap.org recently interviewed him. Check the story here [kerneltrap.org].

Matt Dillon's commit bit was removed before...

[stienman]

KernelTrap [kerneltrap.com] had an informative interview with Matt Dillon [kerneltrap.com] last year about this time, in which he talks of many things including how he got into FreeBSD, what he contributed which convinced the core team to give him commit capabilites, and the friction in earlier years (from his point of view), along with the tidbit that he did lose his commit priviledges previously for a period of time.

It appears (from his perspective) that he did a great deal of work on the VM portion of FreeBSD, but it was (as asserted by members of core) primarily maintenence and bug fixes.

It does seem to give some view of what Matt is like.

-Adam

See Slashdot [slashdot.org]. See Slashdot Google [google.com]. Google, Slashdot, Google!

The flamewar is here:

[otuz]

It seems to have started when Dillon made a clever hack and people got arguing over API problems:

The problem [freebsd.org]
The solution [freebsd.org]
NOT another solution [freebsd.org] ...

The flamewar starts.. [freebsd.org] ...

and continues [freebsd.org].

Re:dillon leaves the FreeBSD project

[dagooncrn]

Yup, he didn't get to core team in 2002 election [freebsd.org]

==vv New Core vv==
rwatson (171)
imp (157)
peter (151)
murray (128)
markm (119)
jhb (99)
grog (88)
wes (85)
kuriyama (83)
==^^ New Core ^^==

brian (79)
matusita (72)
dillon (64)

Arrogant, opinionated, COMPETENT jerks allowed

[Eric Green]

I've been acquainted with Matt Dillon's work for over 15 years now. We're talking about someone who wrote a "C" compiler and library in a week once "just to prove he could do it", for cryin' out loud! Frankly, one Matt Dillon is worth four or five of the rest of the BSD development team. I don't care how arrogant or opinionated he is, I'd hire him for my team in a second -- I've worked with arrogant, opinionated people before (hmm, I've also looked in the mirror before :-), and I don't have the slightest bit of problem with them as long as they're *RIGHT*. Just give'em the module definition, ask'em to produce documentation on what they're going to do, and then once that's done, turn them loose to do it. It works. Been there, done that.

On the other hand, Matt is not, and never has been, indispensible to the FreeBSD project. His biggest contribution probably has been cultural more than anything else -- he was working at UCB back in the "real" BSD days and knows how "it spozed to be". I suspect that doesn't make him popular with some of the (relative) newbies who want to add lots of features and stuff -- Matt's code has always been stripped down, clean, and fast as hell (if not always the most elegant or user-friendly code in the world). If the FreeBSD folks got tired of him carping about "the BSD Way", their loss... but it's not going to cripple FreeBSD by any means.

-E

Re:FreeBSD != Linux

[j-pimp]

FreeBSD is a failure.

So lets look at some facts:

• Yahoo is powered by freebsd
• Hotmail was (is still?) poewered by FreeBSD
• My laptop is powered by FreeBSD
• KDE works great and GNOME ain't to shabby [freebsd.org]
• My laptop is powered by FreeBSD. I switched from OpenBSD because KDE3 wouldn't compile.
• Sun and FreeBSD have come to some kind of agreement so FreeBSD will be distributed with java once sun and java figure out what the agreement is that there lawyers made. Until then all you have to do is download two tarballs, stick them in the righ place and install like any other port.

No its not cool. No its not the place to be for games or embedded systems with GUIs. However, it's great for web/database type things and other places you need scalibility because alot more kernel parameters can be set on the fly than in linux. Also it does have its share of wizbang. Background fsck! Yeah thats boys and girls. System boots up after power failure or kernel panic, fack the / partition, which of course is small and does not contain /usr, /var or /tmp becasue your a competent sysadmin from all those years of slashdot reading. It mounts the other partitions dirty and does fsck while the system is up and running. No its not perfect, no I wouldn't do it on a server yet, but its available in 5.0 and by the time 5.2 comes out (the ok 5.x is ready to take on your super mission critical 5 9's project release) It will be, and its easy to turn off. However, it only failed me once and no data was lost. The point however is there is some innovation going on in FreeBSD land and sure its playing catchup with linux in certain aspects but every damn OS is playing catchup with every other fucking OS out there in some regard. Can you find me one super fast, super GUI, super CLI, Everyones using it OS?

Re:dillon leaves the FreeBSD project

[kraksmoka]

Its great that you found a way to accurately measure how many users there are for any freely distributed software

i call it: AWSTATS [sourceforge.net] .

yes, that's right. it reads weblogs really nicely and only needs old versions of perl to run :)

funny how site traffic ( 50k + hits per month) provides a very accurate picture of what the world runs on their machines. i get about 30 percent xp, me and 15 win2k and NT then maybe 3 win95, 4-5 mac, and a smattering of different linux and bsd distros. a few people here and there with lynx or even someone on a solaris machine surfin the web. i wouldn't say that linux has more desktop share than apple. servers, yes, more linux servers than apple ones, but not desktop users. just how it is.

Quick summary.

[fidget]

(oversimplified, as all summaries must be)

Matt has a holy cow, and has been acting
like a USENET-troll. It's far too easy to send nasty email, still, people.

Reading all the relevant threads in the freebsd lists, here's a quick run-down, with names (other than Matt's removed)

Matt: I have a patch
freeBSDer1: better to fix the ipfw ABI issues (which are in the update log anyway)
Matt: This is bullshit! I'm going to core.
freeBsder2: I have a non-reboot way to fix the prob with an environ var.
Matt: I don't want any !*&&!@ env var! I'm going to core.
freeBSDer1: patch is a security risk, *and* send it through the security officer (SO)
Matt: It's not a *&!*@ risk *&*!@. I want it in.
freeBSDer1: SO established the current ipfw defaults, so changes to that must go through him.
Matt: *!&&^@@ You are being unreasonable.
multiple freeBSDers: No, you are.
Matt: *@$#@!!!
freeBSDer1: Matt, if you can't be civil to your fellow developers (as it states in the conventions), then find another project.
Matt: (rails on about how he shouldn't have to re-write someone else's API just because he's unhappy, when his hack fixes the problem. With expletives, insults, and other nonsense)
freeBSDer3: Matt, send the patch to technical review instead of core.
Matt: This is *&#@%! It's already to core. Why should I have to rewrite the API when this hack is fine?

(more of the same follows, with Matt insulting a larger number of developers every time, and getting more personal by the moment)

Looks like Matt had a couple of very bad days, and vented when he shouldn't.

Is it too late for apologies?

Re:dillon leaves the FreeBSD project

[imp]

What Greg forgets to mention was that this action was taken due to Matt's interdeveloper relation skills, not due to Matt's technical skills. FreeBSD has a code of conduct between developers that the core team is charged with enforcing. Without going into the details of this mess, at Matt Dillon's prior request, Matt violated the code many times over the years. Core tried to bring him into compliance with this code of conduct. After a recent incident, core felt that his future compliance would not be sufficient. His failure to comply to that code was causing damage to the project in excess of his contribution. Core felt it had no choice but to remove his commit bit for the good of the project.

A copy of the committers code of conduct can be found in our handbook [freebsd.org]

Warner Losh
FreeBSD core team
Speaking ex-offcia