People keep bringing up CORBA like it's a bad thing, and while I'm sure it had its issues, I'm now kind of interested to learn about it so I can see what it did get right.

Ohhh, no I *have* heard about this. It's the "everything is just objects and methods" view on computing. Doing stuff like remote IPC.

I believe the fatal flaw in this model was that it didn't surface failures adequately. Because remote IPC calls usually fallible + async. But local method calls don't need to be either of those things. Or at least: local calls don't need to handle network errors.

@yosh the IDL interface definition language wasn't the worst idea. The c++ code generators and complex very modular execution flexibiloty around it though was hard to work with. Implementations especially commercial ones had all kinds of problems, like object A calls B calls C but if A and C happen to be living in the same process, it deadlocked 😂
The gap between nice theory and executable praxis was pretty wide. #corba
@hpk @yosh Indeed, it was a pretty good idea, a reasonable efficient, interoperable binary protocol and a language neutral API description. However, the implementations were usually wonky and not as interoperable as hoped.