So we’ll use…what?

“As a base we need a kernel etc. – one of the BSDs should do.”

Because those kernels are more….something than Linux. Or less something else. Less compatible, and popular for sure. But hey, if the BSD license is so rad-tastic it will liberate us from the evils of the Linux Kernel with all its…Linuxy stuff, then ok.

“(Include OSS for audio.)”

Because it’s old and sucks, and will really showcase the audio capabilities of our hypothetical non-Linux Eagle-BSD…the ultimate desktop…WATCH OUT APPLE! I can’t wait to see who starts mixing records on Logic 8 for Eagle-BSD featuring the OSS audio architecture: A quantum Leap forward in sound sub-systems. Take that, Linux! We’ll have OSS!

” We need an application framework, which should be .NET because it…”

…Blows, has memory leaks, couldn’t even be used as a core language for Windows, is only implemented properly *on* Windows, requires a VM, reflects the biases of the posters here, and may infringe on Microsoft Intellectual Property rights if you implement it in Mono– oh yes, and will result in slow-as-fuck apps like all the ones on Linux you hate. Not even “Linux Zealots” like Mono much, but you’ll show them!

“exists, isn’t bad (AFAIK it’s even pretty good) and there are existing applications.”

For Windows. Or did you mean the GPL apps like F-Spot (ghey) and Banshee?

“X11 doesn’t cut it, so create a graphic “layer” to sit on the kernel and provide the necessary services for the .NET framework. ”

Yeah, only a Linux nerd would use X11. Let’s conjure up a “layer” thingie and provide…WinForms? Or DirectX controls? Good luck licensing that.

“If QT provides any help with this, license it and use it.”

Ummm….with yer thingie layer-dealie? Or?? Is that tied right into the Warp Engines?

Fucking amazing. I think the blog owner cited the Dunning-Kruger effect in another post. I hope his readers Googled that. I just thank the Usability Gods that you guys are here to save the Heathens from the fatally-flawed Linux. Whew.

I completely agree with this post. Linux is completely unusable. I use it every day ”

Yeah, zealotry is a problem. Quite a few on this blog. Wow. Talk about cognitive dissonance.

Project-oriented (asset-oriented) storage in some form would be great. There’s a reason IDEs exist … in fact, representing the entire file system in an IDE (with pluggable extensions, yet) would be a heck of a better paradigm that this stupid folder thing.

Of course, third-party installers (ie the whole of the rest of the world) would still insist on the old paradigm, so you’d still need a mirror filesystem that looks like, sigh, the Unix one. Well, Linux still needs a CLI. I guess you need a certain degree of backward compatibility.

Secondly, project-oriented storage. It’s a real pain having to try and force a folder hierarchy to mimic a project. I propose the idea of project-oriented storage with projects and assets rather than folders and files. The benefits to this become clearer when you take into account the ability to reference one asset into other assets elsewhere within a project. Likewise, all tasks/applications that contribute to a project become linked to it so that there is a complete history of changes made and those changes can be baked, altered, or rolled back.

Think of it this way, you don’t have a Word document, you have a text asset that needs to be combined with a logo asset made of vector art that can then be consumed by a printing task. This way, tasks can consume and remodel a project as it deems fit to complete its task.

@Carsten: Pretty much everything Joel has written about UI development would apply, although I have a sneaking suspicion that Kerberos was there before him and has thought more carefully about it afterwards. It’s a useful resource for ignoramuses like me, however.

*Obviously* you’d have to include aesthetics, however difficult that criterion might be to judge. Where would anything marketed by Jobs be without aesthetics? There’s at least one basic issue to resolve asap, which is to choose between pixel-based font-rendering and aesthetics-based font-rendering (Joel again).

I’m a programmer, not a designer, so a couple of other things spring to mind.

You also have to consider Unity. (I can’t think of a better word to describe what I mean.) An obvious counter-example to Unity is the long-standing X issue with cut & paste. Who knows when it will work and when it won’t? It’s not unreasonable to start with the Linux Desktop as a series of inaccurately linked design decisions and then consciously strive to avoid all of them. (From an Apple point of view, Unity seems to be embedded by virtue of Cocoa, so I’d still want a mandatory scripting framework for absolutely everything.)

One more thing which is *not* part of any current desktop, so far as I can see. There’s enough computing power out there, these days, to allow for an UberPaperclip to offer sensible AI-style help. Maybe current HMI work is going in this direction, although I have no information on the subject. It’d be really nice if one could utilise anti-spam techniques in reverse and actually “guess” what the user is thinking — related files or apps, maybe relevant personal contacts or music/visuals, whatever.

It’s not strictly part of the remit, but it would be a fabulous addition. No idea how you’d make it unobtrusive, though…