subtext, blog comments edit

As mentioned in an earlier article, I updated to the latest Subtext and was having email problems. I have the problem solved now so I do get notified when comments and contact form submissions come in.

For those more technical and interested in what happened…

…it was a lot of things conspiring against me.

  1. The Subtext contact form specifically doesn’t send you email if you’re logged into your own blog. It checks to see if you’re logged in and, if so, just skips the whole send procedure but still says “Email sent!” making for a difficult debugging experience.
  2. The whole way Subtext sends email has been changed. There’s an email service that uses email providers that do a do-si-do and an alaman left and somehow email gets pooped out the other end. I have a feeling that something in there changed without me knowing it, but since I can’t attach a debugger to it, there’s no real way to tell what.
  3. My host seems to require authentication for SMTP now. I don’t know how I was getting emails through before, but it worked in Subtext 2.1.0.5 and after switching to Subtext 2.5.2.0, I was forced to set new SMTP parameters to handle the authentication. This also ran me into the fact that emails from Subtext, by default, come FROM the user sending the comment form. The new authenticated SMTP server at my host doesn’t like that. Trying to figure out the magic combination of parameters through trial and error was especially trying because…
  4. The logging around failed email sending in Subtext is lacking. It may be that there’s just nothing coming back as an exception or something down the stack is getting swallowed and not logged, but there was no indication anywhere about a failed email send.

Anyway, if you upgrade and run into the “email isn’t being sent” issue, first make sure you’re logged out. Log out of both the HostAdmin and the blog proper. If it’s still not working, THEN look at your config settings.

halloween, costumes comments edit

We had 16 more trick-or-treaters this year than we had last year and the most popular time to visit was between 6:30p and 7:00p, which is earlier than the last couple of years. We had seen a trend where kids were coming out later, but Halloween was on a Sunday so I’m thinking the kids had to be home earlier on a “school night.”

Here’s the graph:

2010: 259
Trick-or-Treaters

And the cumulative data from this year and the other years we’ve tracked:

  Year
2006 2007 2008 2009 2010
Time Block 6:00p - 6:30p 52 5 14 17 19
6:30p - 7:00p 59 45 71 51 77
7:00p - 7:30p 35 39 82 72 76
7:30p - 8:00p 16 25 45 82 48
8:00p - 8:30p 0 21 25 21 39
  Total 162 139 237 243 259

As mentioned, Halloween was on a Sunday and we did the two giant bags of candy from Costco like we did last year. We had a little left over, maybe a third of a bag, so it seems two Costco bags is the magic number.

I had intended to decorate more than last year, but I ended up with only putting out the projector again. We’ve had a lot going on lately so I had to skip on the heavier decorating. I was pretty pleased with my costume this year, though, which was Sherlock Holmes:

Travis Illig as Sherlock
Holmes

I made the hat, coat, and vest myself. I’m pretty pleased with how it turned out. I did a decent job matching up all the plaids at the seams and got several compliments, which is always nice when you put in a ton of effort.

Over the past couple of years I’ve noticed a lot more of what I’ll call “Halloween Bullshit” going on. I ended up tweeting about it as it was happening, and while it’s not like “Holy crap, this is a huge problem” or anything, people are doing stuff that, well, I feel is probably less than in the spirit of the holiday.

  • If I see a big group of kids leave your house to go trick-or-treating, you’d best leave someone at home to hand out candy. I watched as a huge family with at least five trick-or-treat age kids left the house, shut off the lights, locked the doors, and went out to harvest candy without leaving anyone at home. That’s leeching, folks, and no one likes a leech. If you’re going to go out into the neighborhood and take candy, have the common decency to leave someone at home to hand out candy, too.
  • If you couldn’t be bothered to wear a costume, you shouldn’t bother to trick-or-treat. I see this mostly in the older kids who probably shouldn’t be out trick-or-treating in the first place. Plenty of jeans + sweatshirt + flip-flops showing up with a pillowcase demanding candy. The “I’m a student” or “I’m a regular person” line to try and justify it isn’t clever or funny, it’s just bullshit. You don’t have to have some elaborate get-up, but at least put in some effort or just stay home.
  • If your kid can’t walk and/or talk, they’re not old enough to trick-or-treat. I’m not stupid. I know the candy’s for you. That kid doesn’t even have teeth with which to chew this peanut butter cup. Go buy your own damn candy and stop freeloading off the neighborhood. We actually had two ladies - not in costume - wheel a toddler - in costume - up to the door in a stroller and hold out two bags for candy. Really? There’s only one kid here. Two bags? You’re not even trying to hide it. (There was a very specific demographic of people who did this. I won’t comment on exactly what that demographic is, but experience in the last couple of years says it’s definitely this demographic that thinks this sort of thing is OK.)

Looks like Halloween is on a Monday next year so I anticipate attendance will be down slightly, and possibly shifted into the earlier times the way it was this year.

dotnet, process comments edit

I’ve spent the majority of my recent career working on fairly complex systems. Integrated services across disparate business units with different data centers. Full multitenancy for SaaS hosted solutions (remember all that Ray Ozzie hubbub back around MIX 07?)… that can also be deployed on-premise for larger customers who want to customize more than the configurable abilities in the hosted environment allow. It’s not The Most Complex System Ever, but it’s not what I’d consider your entry-level project, either.

The thing is, a lot of time gets spent doing things like…  Trying to pull configuration out of XML files and into a central service-based configuration store. Localizing for multitenancy where there’s not just culture fallback to consider but also default values and per-tenant overrides (sort of content management-ish). Correlating logs that run from the end user all the way through to the [disparate business unit] back end systems and back to the user.

Where are all the tools that are supposed to support larger apps and more complex use cases like that?

Based on my personal views and having no scientific data whatsoever to back it up, here’s what it feels like is going on:

[Where I think time is getting spent (click to
embiggen)

There seems to be a ton of stuff trying to get people “just starting out” up to speed… but once you get past a web site that uses LINQ to SQL or whatever to display products out of the Northwind database, where’s my tooling? Where are the solutions to the distributed configuration problem? Where’s the solution to getting resources out of .resx files? Where’s all the multitenancy support? How about even the ability to change the web.config without restarting the application?

I just feel like I spend a ton of time on infrastructure, something we all know Product Management doesn’t want to pay for because it’s not something you can see or click on, and not much time on more visible features. And I’ve mentioned stuff like this before.

Venting? Sure. But am I alone in wondering where this stuff is? I don’t think so.

media, music comments edit

My iTunes library is currently around 160GB, much of which is stored in Apple Lossless format. I also have podcasts, TV shows, and movies in the library. I have a 160GB iPod Classic. I want all of it on the iPod.

Luckily, iTunes has this nifty option “Convert higher bit rate songs to 128 kbps AAC” which will, on the fly, convert the Apple Lossless stuff to smaller (and lower quality) files so it all fits. This is fine for my iPod since my major use case there is middle-of-the-road earphones at work and podcasts anyway.

The problem is, syncing that much and doing the conversion literally takes days. And if iTunes crashes in the middle… you basically get to start over.

What I’d find is about a day and a half in, I’d get a little dialog that would pop up and say “iTunes has stopped working.” From there, if I tried to start things running again, I’d get “Verifying iPod…” and nothing would happen. (I figured out how to get past the “Verifying iPod” message, but it still restarted the sync from the beginning.)

After a long process involving syncing small blocks of the library one at a time onto the iPod and which I’m now referring to as The Great iPod Sync of ‘10, I found the culprit: One. Single. Bad. Track.

Windows Explorer showing ONE TRACK not
working.

See that track where Windows can’t pick up the metadata information? Of the over 14,500 tracks in the library, iTunes encountered this one bad track and died. Blammo.

(I didn’t discover it earlier because I wasn’t compressing and apparently iTunes will just blindly copy the bad track over without checking.)

Once I removed this track from the library, everything synchronized fine.

So… if you’re finding iTunes crashing in the middle of your sync over and over… go look to see if you have a bad track.

process comments edit

I’m a contributor to the Autofac project and recently they switched their repository from Subversion to Mercurial which means I have to learn Mercurial. I’ve been working with it a little and the workflow process is… not quite as intuitive to me as Subversion, but then, it’s a pretty big mental shift and I’m still working on it.

These sites have helped me get up to speed and start understanding a bit better. Maybe they’ll help you, too:

Got other good ones? Leave ‘em in the comments!