I’ve been using Himalaya to give my agent swarm email comms, which is currently their main method of communication after my quick-and-dirty messageboard convergently evolved until I realized this was ten times more work than just giving them each an email, which also opened up lots of doors for autonomous work that requires an identity.

The email provider I’m using, which is the rinkydink shared host I’ve used since 2010 but that gives me lots of space and has some historical sites and honestly I just can’t be bothered migrating anywhere else even though I’m way overpaying for it, was silently rejecting outbound emails while giving every indication of happily accepting. They went around to other emails on the same server no problem, but never out. My agents were kind enough to bisect and realized that the issue was a client EHLO of 127.0.0.1, which Himalaya email hardcodes (can you don’t?). Any other value, even a bogus one, makes it out; loopback doesn’t. And, sadly, Himalaya hardcodes 127.0.0.1. Which seems odd, given that’s supposed to be an identifying hostname, but I build open source, I get it, some user goes “hey it’s super weird you did this” and you’re like “yeah my bad but could you not be so annoying about it.”

So I’d submit a patch, which would be super small, and make Himalaya more standards compliant, and mean I don’t have to give my agents custom binaries, except I’m contractually prohibited from open source contributions for as long as I’m employed by Apple, which I signed willingly, and failed to realize how much pain in the butt it would cause me.

So I can’t submit a patch. So I’m complaining here and hoping someone who did not make a deal with a big company to tie their own hands. I would distribute my own binaries that I use but they’re big and that’s sketchy without accompanying source so I won’t. But anyway, if someone patches Himalaya to get that fixed, I’d be a happy camper. Hey maybe I can get one of my agents to post it on one of those rent-a-human message boards…