"Dumb strict relays, smart strict clients."

Bad implementations are unavoidable given that it's always a work-in-progress. What we should do, is be extremely strict, report/log errors, and at every instance sort things out and fix the error at the root.

Being too lenient makes control flow obnoxious and complicated where it doesn't have/need to be.

Reply to this note

Please Login to reply.

Discussion

you sounds like Charlie Browns teacher.

Sorry, don't know that. Why I'm saying "strict" is because it is in everyone's best interest. Specs get more accurate and precise in expressing what they can and cannot so. Servers/relays can be precise/minimal as possible. Clients can be exact and minimal and without excessive error handling.

basically like a bobblehead doll on a dashboard which keeps making unintelligible sounds out if its face hole every time we go over a speed bump.

Are you describing Charlie Brown's teacher?