Try entering this into Claude or ChatGPT:
Here's a screen shot of the Claude response.
A while back Matt was giving me grief, in a friendly way, about how scripting.com still uses http addresses. I could switch over, but then all the images and included files posted before 2014 or so would break. The minor gain in security on a site that doesn't ask for any private information, is totally not worth throwing out all the work I did on a site that actually has historic importance is just a bad deal. It would be a solving a problem no one but Google has (and it's not even clear what that problem is, and why I should care). There's a principle here too — letting one company dictate to us how the web works, well I got into the web to get away from that.
Anyway, the reason they still use http in a place where one expects https is apparently is the same reason. It would break things that they don't want to break. I'm not suggesting they change it, but somewhere in my codebase somehow the http addresses are getting converted to https, and I haven't (yet) been able to track it down. I'm pretty sure it's a bug I unknowingly introduced.
PS: When I'm calling through the API, I get back a record that has a different guid from what's in the feed. Seems like the API and the feed should be in agreement. This is the code that gets the post record. My guess to get them into agreement, I'm going to have to hack this, changing https to http. And there is the reason they can't fix this, and just have to live with this mess. I think overall the people who manage the feed and the API are doing a pretty great job, btw. You have to know I wouldn't say that if I didn't believe it.
PPS: I reported the problem once it was fully diagnosed, on the WordPress repo for Calypso.
