I can only conclude that PeeringDB is increasingly vibe coded (in the fullest meaning) because they keep releasing busted code that clearly has not been tested. Today they email me and a lot of other networks with a non sensical email that tells me to do something that does... nothing. https://github.com/peeringdb/peeringdb/issues/1936 I can sometimes understand the desire to use AI tools, but seemingly none of this stuff is actually going through even basic testing in a staging environment, so frustrating for a service that absorbs quite a lot of sponsor money from orgs
one honk maybe more
tychotithonus@infose..
replied 25 Mar 2026 16:14 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/gNjKMV9xP4RtVMRfDV
@benjojo Hopefully they'll catch on that ad-hoc testing doesn't scale with such volume, and make test-driven development a hard requirement. If they don't sharpen that saw, the users will take the brunt of the testing burden.
benjojo
replied 25 Mar 2026 21:32 +0000
in reply to: https://infosec.exchange/users/tychotithonus/statuses/116290699382370222
@tychotithonus there isn't really volume though, the core database is like 100MB with indexes and the site has a pretty fixed number of features. It could be a checklist and a working staging environment, it doesn't seem that either of those exist
tychotithonus@infose..
replied 25 Mar 2026 22:05 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/z1yL3J6gb5Vdg78vbD
@benjojo Ooh yes, totally agreed about that gap -- and likely one that predated their (potential?) LLM use, and now "just" magnified by it. (The "volume" I was referring to was the volume of changes that LLM coding -- for good and/or ill -- produces.)
jamesog@mastodon.soc..
replied 25 Mar 2026 18:11 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/gNjKMV9xP4RtVMRfDV