I'm starting the iocaine 3.6 release process by first deploying it on Eru (along with a current Nam-Shub of Enki). Will cut a release after a few hours of looking for trouble.
0 Boosts
0 Likes
Oooh, it might be an ai.robots.txt thing. Infra uses iocaine 3.6.0 with ai.robots.txt 2.1, while NSoE's using iocaine 3.5.1 with ai.robots.txt 1.51.
Yep, another way too generic entry in ai.robots.txt, sadly.
That means patching both NSoE and QMK. Thanks, NSoE test suite for catching this! Tests are good.
For iocaine 4, I will adjust the ai.robots.txt rule: rather than using the list as-is, and handling special cases after, I'll filter the list first, removing problematic elements. Said problematic element list will also be configurable.
Currently, this means 3 entries: code, dot, nbot. I have no data on the latter two, so just dropping them is fine. For code, I'll have a separate check, since I have enough info to identify it. That check will be part of the ai.robots.txt ruleset, but will be performed separately.
Huh. Nam-Shub of Enki's test suite fails during a Nix build, but only when compiled in context of my infra config. Running
nix build .#within NSoE works fine.