Interestingly enough, the "N" in "GNOME" was due to the fact that it was initially based on CORBA, and used a C-language ORB called orbit (IIRC).
If someone older than me would like to chime in and add some details, that would be lovely.
Re:When will web browsers be complete?
by bluGill on Wednesday October 28, 2020 @09:59AM
(#24918774)
KDE and GNOME agreed in the early days that both desktops would be based on COBRA and use standard object names so the two would be compatible.
After fighting COBRA for a while KDE said this is too complex/slow and wrote DCOM (I might remember this name wrong) which gave them the parts they wanted without the overhead. GNOME stuck with COBRA longer, but I get the sense that they half agreed and were just hoping they could find a solution without abandoning COBRA (GNOME also had more corporate backers even in those days so there was probably non-programmer architects demanding they make it work). Evenually both agreed that they needed something more lightweight to solve the desktop needs and DBus came along.
CORBA, not COBRA :)
Also, iirc, KDE's thing was named DCOP
The issue with KDE was that Qt was nonfree so there was also a few efforts to replace Qt with something else.
I worked for a company whose software was/is based on CORBA and let me tell you, that thing really smells like microservices-before-it-was-cool.
You had a naming service, and each component would connect to the naming service on startup and resolve the endpoint for the components it needed.
And each piece of software had a component called ORB (object-request-broker) that was used to represent remote stuff (remote objects/endpoints). You could invoke method calls on this object (kinda/sorta, i'm making it simpler here) and it would forward the call in background, hiding the networking/serializing/deserializing details.
That whole things kinda reminds me of protocolbuffers, etcd and microservices. By using a cluster manager (Veritas Cluster Manager) you had a service groups, something that looked like a pod, in a way.
-----------
It's been a while, but I've started thinking that there's nothing really new. The problems we're facing are pretty much always the same, it's just that the solutions improve on different areas. Kubernetes for example helps a lot on the scaling problem and the standardization (both ops team and dev team have a common lingo, the kubernetes object model -- pods, deployments, services, requests&limits etc).
EDIT:
Just skimmed the article... Yup, DCOM was a competing alternative to CORBA.