Posted by KA9NWM
For most of the first week of July, I stepped away from the app itself and spent nearly the whole stretch on helioclock.com. It’s easy to underestimate how much time a real product website takes until you’re actually building one page by page.
I built it on WordPress, hosted on Pressable, with GeneratePress as the parent theme and a custom child theme underneath it. Ten pages went up: Home, Features, Screenshots, Applications, SKUs, For Hams, Contact, About, and a Documentation page that embeds the app’s own help file directly. I matched the whole visual design to the app itself rather than picking some generic theme palette: the same dark background, gold, cyan, and green that Helioclock’s interface uses, with the actual fonts self-hosted rather than pulled from Google’s CDN.
Most of the pages went in as native Gutenberg blocks rather than raw HTML, which took longer up front but means the text is directly click-and-edit from here on, no digging through markup every time I want to fix a typo or update a number. I made exactly two deliberate exceptions to that rule: the multi-button hero rows, and the Documentation page’s sidebar navigation and iframe, both of which are structural page layout rather than prose, and don’t belong as editable blocks.
Two real bugs turned up while I was building the site, both worth mentioning because neither one had anything to do with the website itself. First, the kiosk snapshot feature (the one that lets you grab a live screenshot of what’s actually showing on a unit’s display) was failing silently. Chrome’s remote debugging port wasn’t included in the kiosk’s own startup service, so there was nothing on the other end for the snapshot tool to talk to. Once that was fixed, a second problem showed up right behind it: the screenshot payload itself was too large for the debugging protocol’s own message size limit. Switching from a raw PNG capture to a compressed JPEG solved it cleanly.
Second, and more consequential: NOAA quietly deprecated the solar wind data endpoints Helioclock had been using since the very early days, right around the end of April, and I hadn’t caught it until I was checking the app was showing sensible data while building screenshots for the website. The old URLs stopped serving current data without ever returning an error, which is the worst kind of failure, the graph looks like it’s working, but the numbers underneath it have gone stale. I swapped over to NOAA’s replacement endpoints, which also came with a changed data format (a flat, newest-first array instead of the old chronological one), so the parsing logic needed real changes, not just a URL swap.
That episode caused me to add a new element to the design: A data fetch health checker to let you know if a data feed stopped.
By the time the first week wrapped up, the site was live with real content, matched to the app’s identity, and I’d caught two bugs I might not have found for a while longer if I hadn’t been staring at real screenshots for days straight.
73,
Anthony, KA9NWM