Comment by linguae

I could think of many examples, but I'll talk about the top four that I have in mind, that I'd like to see re-evaluated for today's times.

1. When Windows Vista was being developed, there were plans to replace the file system with a database, allowing users to organize and search for files using database queries. This was known as WinFS (https://en.wikipedia.org/wiki/WinFS). I was looking forward to this in the mid-2000s. Unfortunately Vista was famously delayed, and in an attempt to get Vista released, Microsoft pared back features, and one of these features was WinFS. Instead of WinFS, we ended up getting improved file search capabilities. It's unfortunate that there's been no proposals for database file systems for desktop operating systems since.

2. OpenDoc (https://en.wikipedia.org/wiki/OpenDoc) was an Apple technology from the mid-1990s that promoted component-based software. Instead of large, monolithic applications such as Microsoft Excel and Adobe Photoshop, functionality would be offered in the form of components, and users and developers can combine these components to form larger solutions. For example, as an alternative to Adobe Photoshop, there would be a component for the drawing canvas, and there would be separate components for each editing feature. Components can be bought and sold on an open marketplace. It reminds me of Unix pipes, but for GUIs. There's a nice promotional video at https://www.youtube.com/watch?v=oFJdjk2rq4E.

OpenDoc was a radically different paradigm for software development and distribution, and I think this was could have been an interesting contender against the dominance that Microsoft and Adobe enjoys in their markets. OpenDoc actually did ship, and there were some products made using OpenDoc, most notably Apple's Cyberdog browser (https://en.wikipedia.org/wiki/Cyberdog).

Unfortunately, Apple was in dire straits in the mid-1990s. Windows 95 was a formidable challenger to Mac OS, and cheaper x86 PCs were viable alternatives to Macintosh hardware. Apple was an acquisition target; IBM and Apple almost merged, and there was also an attempt to merge Apple with Sun. Additionally, the Macintosh platform depended on the availability of software products like Microsoft Office and Adobe Photoshop, the very types of products that OpenDoc directly challenged. When Apple purchased NeXT in December 1996, Steve Jobs returned to Apple, and all work on OpenDoc ended not too long afterward, leading to this now-famous exchange during WWDC 1997 between Steve Jobs and an upset developer (https://www.youtube.com/watch?v=oeqPrUmVz-o).

I don't believe that OpenDoc fits in with Apple's business strategy, even today, and while Microsoft offers component-based technologies that are similar to OpenDoc (OLE, COM, DCOM, ActiveX, .NET), the Windows ecosystem is still dominated by monolithic applications.

I think it would have been cool had the FOSS community pursued component-based software. It would have been really cool to apt-get components from remote repositories and link them together, either using GUI tools, command-line tools, or programmatically to build custom solutions. Instead, we ended up with large, monolithic applications like LibreOffice, Firefox, GIMP, Inkscape, Scribus, etc.

3. I am particularly intrigued by Symbolics Genera (https://en.wikipedia.org/wiki/Genera_(operating_system)), an operating system designed for Symbolics Lisp machines (https://en.wikipedia.org/wiki/Symbolics). In Genera, everything is a Lisp object. The interface is an interesting hybrid of early GUIs and the command line. To me, Genera could have been a very interesting substrate for building component-based software; in fact, it would have been far easier building OpenDoc on top of Common Lisp than on top of C or C++. Sadly, Symbolics' fortunes soured after the AI winter of the late 1980s/early 1990s, and while Genera was ported to other platforms such as the DEC Alpha and later the x86-64 via the creation of a Lisp machine emulator, it's extremely difficult for people to obtain a legal copy, and it was never made open source. The closest things to Genera we have are Xerox Interlisp, a competing operating system that was recently made open source, and open-source descendants of Smalltalk-80: Squeak, Pharo, and Cuis-Smalltalk.

4. Apple's "interregnum" years between 1985 and 1996 were filled with many intriguing projects that were either never commercialized, were cancelled before release, or did not make a splash in the marketplace. One of the most interesting projects during the era was Bauhaus, a Lisp operating system developed for the Newton platform. Mikel Evins, a regular poster here, describes it here (https://mikelevins.github.io/posts/2021-07-12-reimagining-ba...). It would have been really cool to have a mass-market Lisp operating system, especially if it had the same support for ubiquitous dynamic objects like Symbolic Genera.