For once I was true to my word, and I launched.
In no real order, some takeaways thus far:
A lack of forward progress and an influx of complicating factors set back internationalization a bit this week. The good news is that the frontend work is pretty much done (minus a few bad pushes in which I broke the draft recipients modal for a solid hour or so); the bad news is that I hate Django’s backend internationalization model, which relies on a lot of relatively outdated tools.
I’ve expanded the script I talked about last time to handle backend work, and it mostly works (except some issues with newlines, which I think is a self-inflicted wound.) I forgot a few things in the initial corpus of strings I identified:
The first one’s not a big deal; the second one means I have to do a bit more architectural futzing. This is perhaps my least favorite stage in the product cycle: the end is kind of in sight, and there’s a janky thing I could rush out the door and deliver to folks, but there are a lot of landmines in the way.
A previous version of me would have been too cautious, and delayed shipping indefinitely in favor of getting the thing perfect. Now I think I’m on the other side of the spectrum: I am too eager to ship, and that results in a lot of paper cuts and pain along the way. I’m trying to reach a little bit more balance with this, and so I’m taking these setbacks in stride. After all, this is pretty much all uncharted territory for me, and on net its been less painful and more fun than I would have otherwise anticipated.
Jon Yongfook tweeted about my perennially favorite phrase, “engineering as marketing”, earlier this week. This is a subject that I am very sensitive to, as someone who:
In general, I’ve shied away from this sort of deliverable: the closest I can justify here is, well, rebuilding a marketing site, and then at least I’ve got data (and SEO gods) to back me up. The tweet reminded me of an evening project I whipped up a year or two ago, though, which was an API cost calculation tool.
I haven’t run the numbers, but this has received…probably one thousand lifetime pageviews and <10 conversions? Probably not worth it.
But sunk costs are sunk costs, and I’ll probably spend an hour cleaning it up, updating the costs, and re-launching it. (I like Zach Holman’s take on launching — launching something only once is for suckers.)