Write On Top

I’ve been kind of away from writing lately, mainly because I’m still trying to get something to bite work-wise. Eventually, I’ll get into a comfortable routine again. Maybe if I could find some prompts, like what I did for National Journal Writers’ Month the past three years, I’d be more prolific unprompted.

I have been trying to get together with the Hatch Norfolk folks. I see the things they’ve spawned, and am enthused. Some of the discussions I’ve had have made me think more of Bob Zubrin, and his proposed approaches to a Mars trip.

One of the things he really emphasized in The Case for Mars was that going big was a recipe for disaster. He used examples from polar exploration to drive that point home. In IT these days, unfortunately, “go big” seems to be the answer to everything.

I disagree wholeheartedly.

Use native facilities wherever possible. Actually develop through your requirements, and make architectural decisions based on those requirements, not on custom, or sales’ whim. Approaching a problem in that way will eventually get you where you set out to go. It may not be the way you originally whiteboarded it.

That’s okay!

Unfortunately, as I’ve been trying to get my business going, I’ve had to reflect back, and see when I’m about to make the same sorts of unwise decisions I’m trying to help people avoid. Figure out what you need to do, then choose the tool that will meet that need the most efficiently. I was reminded of this pretty quickly trying to figure out Postfix SMTP authentication for the first time in years. Yes, I could have figured it out, eventually. Yes, dovecot provides a mechanism to make it work. Yes, it’s still a royal pain trying to get it to work. Yes, I’d still need to buy certificates so my MUAs wouldn’t bitch.

Or I could just throw a few bucks towards a competent provider, get more reliability than I could ever hope to match, update the MX records, and move onto something else. (Resisting the urge to snark about doing it $whoever way.)

Other stuff from my interactions recently:

  1. I guess Woz is coming to speak, locally. I watched some of his stuff when colleagues were borrowing some of my stuff for the 5th HOPE conference where he was a featured speaker. If tickets aren’t outrageously expensive or hard to get, I’d go. The stuff he did with the Apple ][ was legendary. He didn’t really work on the original Macintosh; if you Google around a bit, you can find numerous quotes from the 80s and 90s where he hated on teh Mac. With the Classic OSes (System * and OS 8/9), I can understand where he was coming from. Apple’s NeXT buy seemed really unwise in 1997, but I think it played out pretty nicely.
  2. For a long time, I was very much against the use of databases. I’ve come around on that, though. For data that is infrequently written, but frequently read, it’s tough to beat a database. Whether that’s MySQLariaDB, or MongoDB, or Oracle, or SQL Server, or whatever, the data is stored, access is provided to it via a predictable API, and it’s not bound up in some proprietary data blob. There’s a reason Microsoft deprecated Access/JET.

So, more, more often, but I have an appointment now.