I mentioned this in yesterday's daily report (which happened to have actually been written about 30 minutes ago). I'm pulling 14 different files instead of the single US CAP alert file, because the library I'm using (noaa-alerts) doesn't appear to support extracting/inferring the state data.
It's currently taking over a minute to run the CAP updater. I'll bet it would run a lot faster if it only used the US file.
At this point, it looks like the hand-off from the National Hurricane Center to the Hydrometeorological Center is going to happen later than previously predicted, but I still want to make sure that Cyclocane will be as ready as possible for when the hand-off does occur, so that coverage can continue uninterrupted.
The real todo item is that I need to break out the updates into individual pieces for everything, but right now, the CAP alerts are what's most brittle, so they're the priority.
So prepare to be amazed/horrified. Cyclocane pulls data from something like 10+ different sources at this point. The way things are currently written, if any of these sources fail to respond or there's an error parsing the data or whatever other unforeseen consequences occur... the entire site update is aborted.
But since this actually happens fairly infrequently, it hasn't been a priority. I hate to complain about it, because the noaa-alerts gave me such a huge jumpstart on the scraping of CAP alerts, but it seems to be fairly fragile at this point. It's probably more so the fault of whoever's producing the CAP alert data, but the point is that one tiny breakage currently causes the entire site update to stop.
And of course, if I do finally break everything into individual components, I'll also have the option of updating some of the sources of data more often, and even scheduling some of them for those data sources where the update times are known in advance.
But the whole point of these pick 3s are to work on 3 separate projects, so here, let me come up with a few more.
Write at least 2 new articles. They can be short. But the point here is to get in the habit of writing every day.
What the hey. Let's go for broke. Or whatever the expression is.
I'm already intending to do something more general with the CAP alerts now that I'm finally parsing them. For background, I've wanted to do something with them ever since I discovered that the source that wickedwx gets its data wasn't the most reliable.
So, after I sort out the national bit for Cyclocane, I'm gonna put all of the alerts on severe weather hq which is currently unused (these days, I'm all about doing something with the domains I've already paid for).
Bonus points for if I make it an Angular JS frontend and figure out how to filter by month and advisory type (I did something similar with wickedwx, but much more by hand).