It’s crazy to think a Raspberry Pi 5 would wipe the floor with a PowerBook G4 laptop in terms of raw performance.

Wish we could have (magically rewritten) native OSX Tiger optimized for open ARM / RISCV laptops. Yeah we have GNUStep but I mean real OSX with all the attention to details this meant 20 years ago.
Aside from the unimaginable amount of work I fear what would break this dream today would be encounter with the modern ad and JavaScript-infested web.

So many websites can bring a  class CPU to its knees and of course ruin the whole visual. Running an old PowerPC Mac today I still a great experience until you meet with the internet.
A center piece of a reimagined Peoples’s OSX Tiger rewrite would have to be some solution to simply replace the web browser with native modules for the most common stuff so you can avoid it most of the time. Not always, just most of the time.

A sort of Reader Mode everywhere but on steroids. But one that is also a Writer. And where sharing privately / publicly is just natural and transparent. A checkbox.

All content types are defined by a set of rules defining what content is. As simple as possible: folders with files and a few json for attributes.

These could be called Collections. You create collections of texts, or images, or sounds and they have attributes / links to each other.

The main idea is that you create content but appearance is ultimately left to the viewer module.

A Collection of .md files is what makes a blog for example. You have dates and other attributes but the viewer client decides how this is laid out based on user prefs among sane choices.

A Collection of .wav files would define a music album. A video series would work the same way.

#Fantasies #RetroFutureComputing
@santi not even enough ram to run modern Chrome on something from the early 2000’s.

OS-X Cheetah technically could run on 64MB of ram. The max you could put on an iBook I think was 320MB.
@bryan I sometimes run modern WebKit on G4 on #MorphOS (on a 800 MHz iBook) and it is so evident that even though it does work it just is a monster.

Typically on MorphOS every thing is lightweight and launches instantly. But then you hit the browser component… it is only really comfortable on smolweb and then it doesn’t make any sense having this engine around that is likely heavier than the operating system itself.
Editors would have  carefully crafted native modules for viewing / editing each type of content and they would absolutely need to be native.

Viewers would be theoretically replaceable but the key of the whole experience would be minimal but good enough default modules that show how it’s done.

Then if someone wants the pure console equivalent or some proxy to make it viewable on an 8 bit computer, or more accessible, it can be done easily because the rules that define basic content types make that easier, unlike interpreting a modern web JS-ridden page.

Of course for a smooth transition you’d have  both the content to be viewed in native in the system and simple web exports for the outside world just showing off how cozy it is.

The temptation would be to make all this with HTML / CSS and base the whole system off it but this would defeat the whole idea of not having a humongous browser engine in the back.

#Fantasies #RetroFutureComputing
Not sure what the underlying sharing architecture should be. Use of distributed P2P systems tend to be a bit hostile in terms of user experience. Kind of the mentality of the associated crowd and often illegal content.

From a user perspective it should be just marking the Collection folder as shared and anyone with the link could view / sync it locally view it with the native viewer for that mime-type and optionally get updates .

It would be simpler and more efficient to have say front end WebDav servers cache the public content folder for serving / syncing. Like one author allowed to PUT but world readable . Also makes sense to have a flattened smol web version of the content for those who are on other systems.

Private / limited sharing would be a whole other world and I am not sure I want to go there yet.

My fantasy is more about public content publishing and making it easy, lightweight, inevitably beautiful leaving no space for advertising. Only content in the same way you can setup a #geminiprotocol capsule but make publishing one click. The way Apple would have wanted it circa 2005 but not in a walled garden.

Ideally each person would his/her own domain and server but more realistically in the way we had Fediverse servers generous people could setup the public servers to make public content easily accessible in a subdomain / folder.

#RetroFutureComputing
To be fair a lot of this is actually doable as prototypes at least. I can’t realistically magically recreate an open source OSX Panther as the underlying base  but I could build some viewer / editor modules a on top of gnuSTEP with all the limitations this implies.

But it’s not like I do anything anyway :-)

Having late accounting issues to deal with and associated anxiety is definitely the reason I am musing about this from 4 to 5 in the morning.

Still who knows if future me feels better one day …

#RetroFutureComputing
@santi Become better soon, all the best to you!
And yes, Panther was nice, but the amount of work and genius that went into it, you won't get done by volunteers anyhow . I'm currently investigating good old WebObjects, which is comparatively small sized. It's organized amazingly neat and all open source implementations only provide a fraction of its functionality.