OpenDoc in particular, or the broader genre of publisher/subscriber frameworks? OLE was "successful" on Windows among the Office apps, but I don't think third parties wanted their apps to be relegated to "components" inside views controlled by the platform owner
Conversation
SOM was the underlying model (if I recall). Shared by IBM and Apple. OS/2 bet heavily on SOM, the whole Workspace Manager relied on it, but the tide was turning hard for MS at the time. I agree about why Pub/Sub model failed. There was an “object” market at one point but meh.
3
3
Yeah I don't think IBM or Apple in the mid-90s were powerful enough for anyone to take their middleware seriously. Windows had its whole own parallel object infrastructure, though, and I think 's answer pretty well sums up why devs wouldn't go for it even on Windows
1
4
Yup! Greg nailed it. He’d know for sure. As a nerd I was attracted to the idea (and OS/2 Warp) and it had cool things about it but while it was promising it never quite landed. COM went a long way (too far even!) but the OLE thing never quite worked well for anyone. Worth trying.
1
4
One of my MS jobs was writing sample code for Cairo dev tools team. So one of the only people to write code on (rather than for) their unreleased OO OS (components to the max!). 95% of the code ended up being adaptor-y/shim-y things to plug two FAR too general components together
2
2
..and it all had to be callable from C because that's where the industry was at the time.
So the other failure reason is that we all went crazy on overly general OO, and the tooling and language support was just vastly insufficient for the aims.
2
4
I am a big proponent of trying out ideas from years passed that failed. Good ideas can fail for any number of reasons and if one can identify those reasons and believe the context has changed then give it another shot.
I don’t believe that can be true today with the app model.
2
4
There’s a fair chunk of it was the Open/Save sheets. The host all presents the sheet and a separate process presents content within it. The picker process can see the FS as the user does but the host app never can.
2
4
Indeed! I architected the iOS version of that (“remote view services”) and was very influenced by OpenDoc… but never felt I grasped the full stoty of its failure modes, which honestly terrified me.
Oh! I didn’t know that. (Please forgive all the typos there)
When I saw how it worked I immediately thought of OpenDoc and the possibilities that opened up with that remote view hosting mechanism. Fascinating stuff.
2