Posts for April 2003

2003-04-02: Pet language peeve

Casualty does not mean death. Casualty means a military person lost through death, wounds, injury, sickness, internment, or capture, or who is missing in action. In other words, casualty means "someone who can't fight any more."

Free hint to news agencies: every time one of your reporters says something like "there were no casualties, but twenty people were seriously injured," you look like idiots. And the casualty figures are not fatality figures, so stop providing fatality figures and introducing them as casualty figures. And when you compare a war to a previous war, try to compare fatalities to fatalities and casualties to casualties, rather than using both terms interchangably leading to numbers that don't make any sense because you're comparing apples to oranges.

2003-04-02: Mozilla roadmap

As many problems as the Mozilla project has had, it continues to impress me every once in a while. The browser is excellent at standards compliance, it actually implements CSS properly and reasonably and is quite portable on Unix systems, and the Mozilla hackers really seem to be trying to do the right thing.

A new Mozilla roadmap was just released that gives me that feeling again. They're moving away from a single monolithic application towards more of a component model, which I think is an excellent idea and something the entire software industry could stand to learn from. Actually doing components in a way that's useful is hard, but not doing components is way worse, and at least components let you lower the footprint of some giant application by turning off the parts you'll never use. As someone who always tries to run the browser-only bit of Mozilla just so that it will be a little smaller, I approve.

2003-04-03: A hole in the wall

This is one of the coolest stories I've seen.

An Indian scientist tried an experiment. He put a computer with a high-speed Internet connection in a hole in a wall bordering a slum in India, running a web browser in kiosk mode, made a trackpad for the mouse and a few other basic controls available, and just watched to see what the kids living in the slum would make of it. No instructions or anything.

The kids taught themselves how to use it, making up their own words for the cursor and other bits of the web browser interface that they'd never seen before. They love it. He's now been setting up similar computers in other places all over India.

Read more about it on PBS's web site. This is one of those stories that makes you feel like there are people in the world who really get it.

2003-04-04: filter-syslog 1.10

Added a -n option that doesn't mail the filter output and instead just displays it on standard output. I needed that today while testing filter rules for an MIT Kerberos v5 KDC log.

2003-04-04: Weekend!

And I'm really looking forward to it, too, even though I got very little done this week. But at least I buckled down this afternoon and made substantial progress on the build template for our Kerberos v5 KDCs; it only needs a little bit of work to finish it up and then I can start on redoing the template for the Kerberos v4 KDCs (for as long as we have to keep running them).

My part of the yearly performance review has been written up, and I've now taken (mostly irrelevant to me, but hey, it's the law) HIPAA training, so the week wasn't a complete loss.

On the agenda for this weekend is a free-flowing intention to do whatever I feel like doing, but mostly likely falling into that category are fiddling around with some INN stuff that can go into the tree after the 2.4 release, possibly working on web page things, possibly working on the grand conversion of to Mailman, and possibly working on other software packaging things.

Oh, and watching the NCAA basketball championships.

2003-04-05: runauth 1.11

runauth is a shell script wrapper around various Kerberos and AFS programs to automate running a program with its own credentials. This is the first public release on my web pages, although I've sent copies of it to various people before. We use it quite a lot at Stanford.

2003-04-06: Weekend fiddling

Another fairly quiet weekend. I watched some basketball and got a few minor things done, and that was about it. Although the main page of my journal at least now validates as XHTML Strict and can now carry the validation icons. I still have some HTML cleanup to do, though, and haven't yet tackled all the other pages.

I've ripped out all the trackback stuff for right now. It seems like an interesting journal community thing, but I'm not really doing this for the community yet and I haven't had time to really think about it and figure it out (and fix up the templates for the additional pages).

In other news, I converted four more INN man pages to POD, and I think I'm close to figuring out the remaining bugs with ovdb and BerkeleyDB 3.x for the INN 2.4 release. And I got a good bit of writing done for the first time in quite a while, so the day wasn't a loss. Just... uninspired. I'm currently playing with ideas for character journals, but haven't gotten very far.

2003-04-08: Jollof rice

From News24 by way of Salon:

[President Thabo] Mbeki said those making war on Iraq contended that they had taken up arms to transform that country into a democracy. The proposition was that democracy could be imposed -- "in much the same way that one can force-feed a person on a hunger strike."

"Presumably the argument is that whether a person ingests jollof rice voluntarily or does so because he or she is force-fed, the fact remains that they have eaten jollof rice."

Mbeki added: "I am not certain that the institution of a democratic system can be approached in the same way that we approach the consumption of jollof rice."

You know, I thought the whole bit of making the world safe for democracy was rather arrogant and egotistical, and now I find myself thinking that really wouldn't be so bad compared to the hubris coming out of the Bush administration these days. At least the phrase "making the world safe for democracy" carries with it the implication that the most that we can do is make it possible for people to make their own choices. As arrogant as it is, it incorporates within the slogan the implicit acknowledgement that the real power lies with the population of the rest of the world.

"Regime change" just somehow doesn't capture that important point in quite the same way.

2003-04-10: Stress and vacation

I still need some work on my ability to identify the best point to take a vacation and actually taking vacation at that point. I do think I'm getting closer, though.

It's difficult to recognize the warning signs of stress and realize that it's time to take a vacation and a general break from things and let my stress levels go down again. I normally handle things just fine for an extended period of time and then suddenly begin feeling extremely stressed over the course of a week or two, and then it becomes much more difficult to last until vacation. On top of that, I find travel inherently stressful, so when I go on vacation when I'm already stressed, it results in an increased short-term burst of additional stress and can be very uncomfortable.

The best sign that I've been able to find so far is "productivity" levels on weekends. I put the word in quotes because it's not quite the right term, but it's the right general idea. If I'm spending most weekends just zoning and feel very unmotivated or unable to work up the energy to do anything, this is a sign that I'm approaching that knee in the curve and should take a vacation soon. If I'm tackling a ton of personal projects on the weekends, that usually means I'm doing fine. But I have to really pay close attention to pick up on this, since I seem to come up with a lot of other reasons for why I'm feeling however I'm feeling.

I think I mostly got it right this time, but I'm about two weeks late. That's better than I've done before, but it could still use some tweaking.

2003-04-14: filter-syslog 1.12

I've just released filter-syslog 1.12, incorporating a patch from Steve Benson to optionally include the hostname portion of the syslog output in the mailed report (more useful for systems that aggregate syslog from multiple systems).

I think it's really cool that I can put software up on the web, without a lot of fanfare, and people not only download and use it but send me patches to improve it. That's so much fun.


I finally managed to stop dragging my feet about writing up a summary of the results and an analysis of the current uses of LIST OVERVIEW.FMT in NNTP and whether we should standardize it. Unfortunately, the results didn't really clarify things a great deal, but they did add some more information. The command is pretty rarely used, as far as I can see. I think it's an odd wart that we're probably better off without, but there's still that edge case of a news reader wanting to know what to expect from OVER responses that I can't quite dismiss.

We're down to the last three issues with the NNTP draft: this one, adding LIST HEADERS to the HDR extension (I think this has pretty much been decided at this point), and what to do about the 48x responses. I think the last one is also nearly resolved, so we should be just about done.

2003-04-18: Mmm, vacation

This has been a very good vacation so far (three more days to go). I finished a book, started another, and did a bit of random e-mail writing and now some work on NNTP standardization, but mostly I've been writing code. I've finished my INN patches to use a helper program just to bind to privileged ports, allowing us to remove inndstart entirely and remove a lot of complexity related to it, and I have a good start on a new way of dealing with both the control message archives and active file maintenance on and the news.announce.newgroups archives.

I'm not quite done with the latter, which I promised to work on this week, but what I'm feeling like doing at the moment is taking another step towards cleaning up the INN networking code. Right now, it's full of #ifdefs to add IPv6 support, and I want to make nearly all of those go away. That's going to require that the INN code just call getaddrinfo everywhere, which is going to require a getaddrinfo implementation for systems that don't support it. And that sounds like it would be rather fun to write.

So I think I'm going to do that next, and then go back and work on the control message processing later.

2003-04-19: TinyNNTP

Today, I finished debugging the basic functionality of my current pet project, a miniature NNTP daemon that only supports IHAVE and just stores all the received articles in a directory and informs a processor program about them. I'm going to use this as the infrastructure underneath a new system for maintaining a net-wide active and newsgroups file, and for doing control message and news.announce.newgroups archiving.

The fun part is that I did this by experimenting with ideas for how to write a good NNTP support library to make writing programs like this rather easy, and had good success there. I have several bits of generic code to put back into INN that can then be used in nnrpd and elsewhere. The command dispatch table, for instance, should work for both nnrpd and innd after some fiddling and let us eliminate some duplicate code *and* make command dispatch slightly more efficient.

Next, I need to write the control message processing module, borrowing heavily from controlchan (although I should be able to simplify controlchan quite a bit, and I want to make it run independently of an INN installation).

2003-04-22: DNS fun

So, very few people can actually see this right now, including anyone reading through a syndicated feed, because DNS for is currently completely hosed. This happened because of rather catastrophic happenings with the servers of my friend who was providing DNS service (things totally not their fault and outside of their control that I won't get into here).

Sigh. I really didn't need something new and unexpected to deal with right now.

In a fit of emergency productivity, I now have new DNS service set up (thank you, Brian!), but of course Network Solutions can't manage to do anything at a reasonable pace, and it will be three to five days before the .org server knows about the new DNS servers. At some point after they finally get that fixed, I'm going to transfer the domain away from them to someone else (what can I say, I registered the domain when there was only InterNIC, and I'm lazy about fixing things that are working).

So all of the mail for all users is currently being queued, and at least some of it will probably start bouncing before DNS is fixed. All of my web pages are off-line, as is my archives. My mailing list services are down. About six moderated group submission addresses are not working. And I spent a lot of time letting various people know.

Kind of not what I wanted to be doing today. Although I managed to also get a lot of work done as well (mostly by dint of staying at work until 8:30pm).

Hopefully, nothing unexpected will happen tomorrow. That would be nice.

2003-04-25: Annoyed

The DNS stuff still isn't fixed. Network Solutions is continuing to shine me on, saying that they're researching the problem and will get back to me as soon as they can. Which at this point I assume means Monday, which means that the DNS situation won't get resolved until at least Wednesday due to propagation delays. It also makes me start to wonder if they're not going to be able to resolve it at all or something.

I am, to say the least, extremely frustrated and annoyed. I don't understand why this is so hard. All I want to do is change the DNS servers for a domain name that I own.


I'm trying hard to not let this ruin my otherwise very good mood, but it's getting increasingly difficult.

2003-04-27: DNS recovering

Well, after not getting any traction with Network Solutions on Friday, I was actually expecting to be down for the whole weekend at least (and was starting to get both rather worried and rather upset), it looks like they actually did push the change through in the Saturday night update. My users started getting mail again late last night, and now three of Stanford's four servers are seeing the right stuff and the actual records in DNS are all right.


That was an unexpected ordeal, but at least it looks to be mostly over now. Which renews some of my enthusiasm for working on eyrie-related things, now that they're actually visible again.

(There will be another outage around May 5th as I move offices, but that one is expected and can be planned for, and shouldn't have any unexpected timing to it.)

2003-04-29: WebAuth 3.1

The latest and greatest version of Stanford's web authentication infrastructure has now been tagged, built, and signed, and all the documentation (at least for first pass) has been written and put up in the staging area on the web. We're all set for the May 1st release so that people can start migrating en masse and get off of the old directory servers.

This is what I've been working on basically non-stop since I got back from vacation, so hopefully I should now start getting some free cycles for other things. I've even written up the draft announcement already, and other people will be able to test the prebuilt packages.

As soon as this release makes it out the door, I'll go add a link on my software page as well.

It's more difficult to do release management for WebAuth than for INN; I can really see the differences between having it be part of one's job and doing it as a hobby. There are a lot more web pages to update, a lot more people have to be handheld through the process, there's a ton of internal documentation and notification to write up and communicate, and of course we have to deal with binary packages where INN has no binary distribution. An interesting learning experience. A lot more of it needs to be automated than currently is.

2003-04-30: Office move

As of next week, I will be moving out of the building in which I've spent my entire career at Stanford, and into a rather less nice office which I'll have to share.


I really don't have a lot of grounds for complaint, as pretty much the entire rest of my parent organization apart from managers have been sharing offices for years, and it was only due to an accident of geography and history that my group hasn't needed to. I do think that it's a real shame the way that Stanford sometimes treats non-academic staff, but at least we do get more space than graduate students, and it's a workable size for a two-person office.

I'm rather annoyed, however, that in the process our group will be broken into distributed chunks even more than now, with three of the people in our group (including the manager) at one end of a large building and another seven at the other end of that building (with another person on the second floor and the last person in a different building, although those last two were expected). I'm particularly annoyed because this was something that we explicitly expressed concern over.

I'm also generally a bit stressed, just because of all the things that one has to keep track of when moving. My computers will, of course, move before everything else and I'll be moving them myself; I wouldn't trust them to movers and I want to coordinate the downtime and DNS updates. But I have to change my address on magazine subscriptions, get new IP addresses, track down all the places in DNS where the IP addresses have to be changed, and box up all of my belongings, all little details that I find stressful and not at all fun.

There are advantages. We'll now be in the same buildings as the rest of our parent organization, which should make intergroup cooperation and communication much easier, and it will be rather easier to drop by the offices of upper management and express concerns. And I get along quite well with my officemate, and there's a good chance that sharing an office will make both of us more productive.

But still, the whole situation is vaguely depressing. Parts of it are just not right.

Last spun 2024-01-01 from thread modified 2020-06-15