So, I'm finding myself irritated with OSS again. Yesterday, I found a bug in postgres' handling of ipv4 addresses. However, it was a peculiar bug in that it only affected a kind of notation that none of the postgres developers use. 

 

So their first reaction was "well that's not even valid." After finding that it is valid -- and explained in the manpage for inet_aton -- and finding documentation on a mailing list from Vixie, they agreed that it "might" be valid.

Now you can imagine how many RFC's are out there on TCP and IP. When they asked me to come up with an RFC that said it was a valid abbreviation (eg. 127.1 vs 127.0.0.1), I couldn't find one.
At that point they lost any interest in the problem at all. If they had said that they didn't have time to code it, or that it might be confusing to people, or it would break the parser, i would have been okay with that.

In fact, that's what I expected.

Instead what I got was "we don't like that notation, we're aware it is okay, and we're still not going to use it."

Funny how two projects I love using -- postgres and openbsd -- are the only two projects I've ever sworn off using. Both of those instances were due to some figurehead/demagogue being irritable and prone to offhanded dismissals of dissent (or even bug reporting). Theo de Raadt, and Tom Lane.

Tom's a nice enough guy, I suppose. However, every time I've suggested anything to him, I get not only a "no," but I get a "and your idea is stupid, too."

Unfortunately, as is the case with OpenBSD, I can't just swear off it. There is just no other database anywhere near as good for the price.

This problem is endemic to the community. I wonder how much of it would be necessary to crush OSS entirely. People can't be lead developers and prone to temper tantrums forever. As OSS becomes more accepted, some of the more vocal, flame-prone types leading projects are going to have to grow up and get some manners. It is almost as if they've been in their "worship" circle for long enough that they expect everyone who comes along to be in that worship circle too. How many ninnies have you seen lauding Theo's recent one-line "slackass!" replies?

It must be going to their heads.

cat death

Work is going to pay for a second dsl line to home. Looks like the network is going to be split again. So, if you can believe this, at my home office (I'm a consultant, remember), I have some twenty machines running. These are split between 3 subnets (each of the RFC1918 nets, and a further six or seven sub-subnets) now, with three different firewalls, and two different paths. Aggregate something like 3mbit down and 2mbit up, with eight static ip's. Yow. Additionally we do all our voice traffic with VOIP, and I don't have a POTS line. Cool stuff.
 
Anyways, the reason I bring it up is I'm going to be migrating to IP6 at home/office. I figure its a notion that's a long time in the coming. We have all the infrastructure we need here, I just picked up an Extreme Networks switch capable of the GigE we need for the Macs, and enough 100T ports to get everyone up and sending large amounts of traffic. Everything (obsd, darwin, solaris 9, irix, fbsd 5.2) supports it, it's higher performance. It just seems like a win all the way around.

Been reading the IP6 books on Safari. That's actually pretty neat. I do find myself wanting to download the damn book so I can read it while I'm waiting at the doctors office. Matter of fact, I'm tempted to write a WWW::Mechanize script to do just that. I suppose I am not supposed to be doing it, but would it really hurt anyone if I just read my copy offline? In general, if I have a network connection, I'm not reading deadtree stuff. And if I'm offline, I can't read the safari stuff. Maddening.

I am going to kill my cats. I've been thinking of leaving chocolate in the trash so the next time they spread it all over my apartment, it will kill them. Or maybe a fire. Or maybe an accident with a deep frier.

failures to disclose and people who fear disclosure

find myself wondering today why more large businesses aren't willing to admit they use opensource software. when i worked at ACS it was always so hush-hush that we had this monstrous postgres database. furthermore, we were using linux. 

i've said a few times that i don't really like linux. this is still true. however, i don't think it's right for companies to be using it as heavily as they are without giving some credit to the movement.
so i'm thinking maybe it would be nice if there were some anonymous submission archive where employees of said companies could contribute anonymous anecdotes about how such and such software was used.
 
seems to me that the gpl says if you distribute software with gpl components, you have to provide the source, and the same rights you were given under the gpl. but there's no component that says you have to be truthful about using it, and if you're not distributing the software, you aren't compelled to make it available. on the other hand, there's no body of researchers out there trying to determine whether anyone is using gpl software either. or is there?

this leads me to believe that a license with such a stipulation might not be such a bad thing. of course, licenses are one of the jihads among programmer folk, so i'm sure somebody has a counter-opinion.

minutiae

life at the aforementioned isp has been interesting. its the first time i've ever been really "full time," versus being a contractor (for almost ten years even). the culture is very different. 

 

i'm working with some very bright people, and i can tell this is going to be a challenge. thankfully, the people i'm working with are real nice, and i've known some of them for years online (on irc).

i've now got three cvs repositories i'm maintaining my code in between work, home, and home-public (eg shared). so much stuff to code, so little time.
 
been doing a lot of postgres stored procedure stuff. never really liked sql from an aesthetic standpoint, but i suppose if i were going to code in sql, i'd choose pl/sql. at least it has some control structures. :-/

my mother and sister coming out turned out to be more taxing than relaxing, that's kind of a shame.

been coding a lot in POE. i think Matt has it right when he says "there are two kinds of perl programs i write anymore: 4-5 line quick scripts, and POE scripts." I thought about this some yesterday, and I realized, I see almost no point in coding in "regular" perl anymore. state machines just make so much sense. i think when people tell me i missed out on college, this is the stuff i would have liked to have done. mostly i've found academia to be stuffy and overly impressed with itself. but this kind of thing is fun. sometimes its practically useful, but mostly I just like programming this way. we shall see in the spring what happens when i get back into school.

my friend Rod has encouraged me to publish a paper on self-organizing networks of natural language parsing sensors (eg botnets that you can "talk to" for monitoring), and has said he'll hook me up with a partner of his. i'm excited about this, but nervous. i've got a lot to say and no real coherent thoughts on it yet.

the navy, predictably, got completely pissed. i told them. months ago. "i am leaving if you don't put a job in front of me." so they show up on the last possible day, offer me $15k less than i'm making, and call it a job offer. riiiiiiiight. maybe in a decade or so when i need something relaxing.

starting to think raising kids might be a more rewarding activity than programming. we shall see. sandy seems supportive either way.