njr 3.0

Last week was very busy; despite my pouring copious hours into it, my research is going slower than I could possibly have imagined. For the current round of experiments I'm using much higher-quality tools than I did last semester, and I sure appreciate not having to fix stupid bugs everywhere I look. Slowness comes from the learning curve—so many tools to understand, so many configuration options to confound, so many places a feature could be inserted. I wasn't able to get any results by last Monday; I'm now working for this Monday. (Mondays are the days everyone in our group sends status reports and results, so it's a natural climax to the week; Wednesdays are meeting days, providing an added incentive to get stuff done™).

To change the topic completely, is Mac OS X's sendmail configuration supposed to work out of the box at all? While poking around I discovered that I had thousands of files in one machine's /var/spool/clientmqueue directory, mostly output from cron jobs and subsequent “can't deliver” notifications, followed by bounces after a few days. The log showed messages like this:

Apr 12 05:01:28 byron sendmail[7070]: h3B4PMFk000538: to=root, delay=1+05:35:46, xdelay=00:00:00, mailer=relay, pri=2731595, relay=localhost, dsn=4.0.0, stat=Deferred: Connection refused by localhost

There's a “/dev/null” in root's .forward, but it appears to be completely ineffective in preventing the pileup. Yay.

The only way to get even local mail delivery to work appears to be to run a sendmail daemon. By default turning on MAILSERVER in /etc/hostconfig causes the server to be open to the world, which I would not like. The undocumented (that I could find) M4 expansion for DAEMON_OPTIONS(`Port=smtp,Addr=127.0.0.1,Name=MTA') causes sendmail to listen on the loopback interface only.

Back to my original question: is this stuff supposed to work out of the box, or is OS X's default mail configuration broken and untested? I use Postfix on my servers and don't use or particularly like sendmail, but for simple local workstation use I don't want to rock the boat.

10:52 PM | Radio
No Comments

Renewed Radio today, so I'll be with you for another year. No time for Pester or much else fun this week, head down on research and errands. I've finally got a viable backup system for all my family's machines (combination of hard drives, tape, and magneto-optical), which is very reassuring. hamton is now down, replaced in its Web and mail-serving tasks (save WebDAV, which is currently broken) by arnold.

More on Monday if I get any research results this weekend.

11:39 AM | Radio
No Comments

Jaeyoon Cho's HanTel is a promising new terminal program for OS X I'd never have found if not for my referrer log. The Web page is in Korean, but the application itself is localized into English. It has a friendly interface, is faster than Apple's Terminal.app, has full color and elegant mouse support (the only native OS X terminal client I'm aware of to support the latter), exposes direct SSH and Telnet connections and even Zmodem transfer via a bundled copy of lrz. About the only things it doesn't have are per-window settings, xterm title support, and drag and drop. All in only 7510 lines of code (smaller than Pester 1.1, 11246 lines as of tonight). The source is licensed under the GPL.

9:44 PM | Radio
No Comments

Rendezvous products are certainly making an appearance. Two I played with briefly today are Hydra, a collaborative text editor, and Rendezvous Beacon, which allows publishing of arbitrary resources via Rendezvous.

9:21 PM | Radio
No Comments

The first public source code release of Sakana is available, and it turns out Brian implemented RSS 2.0 support before I had a chance to. Oh well. Not that there's any information in there yet, but you can subscribe to the Rileys wiki changes if you like.

12:44 PM | Radio
No Comments