Editing a running program…

This reinforces the fact that the directory structure of a file system is completely independent from the files. I.e. you don’t need directories to store files. They exist primarily for the user to be able to browse and organise files.

It’s interesting that hierarchical file systems and GUIs emerge about the same time on PCs. Early PCs including Apple II and MSDOS, didn’t have directories. They add them in 1983, the same year as the release of the Apple Lisa, and a year before the Mac.

What I like about this is that apart from ^, no special class was created, no superclass, no function call, no library, no specifying where the file is stored, or which file. A single character and you have a persisted variable even over the network.

As far as I can tell still prominantly used today for Electronic Healthcare Record systems and other industries like finance.

Got MUMPS? Built in 1966. Automatic persistence of variables by preceding with ^, even over the network. Sets the first name of variable A on computer DENVER: SET ^|"DENVER"|A("first_name")="Dobbs"

I think there is some emergent power when the programming language, file system, and GUI are intrinsically linked.

Swift and C# were designed for pre-existing environments but also were designed to be universal and not tied to their environment.

In other words there aren’t many languages that are closely intertwined with their environment. JavaScript may be the only notable one.

When was the last time a new programming language was created for an entirely new domain? Smalltalk, Self, HyperTalk, and NewtonScript come to mind. Also JavaScript. Any more recent? Most recent languages seem to be universal rather than environment specific.

Int vs Floating Point is topical at the moment. Pop quiz: Which can store more numbers: A) 32 bit Int B) 32 bit Floating Point

Perhaps most importantly, you don’t need to align the camera with the scene. The camera is perfectly aligned so that the background plane fills the viewable rect. It shouldn’t even feel like the user is in a 3D world, even though they are.

But you get textures, shading, lighting, shadows and orthogonal perspective. To me that seems worthwhile without much user pain.

Imagine something like Keynote where all the objects are 3D, but you have almost no control over the z axis. Z axis control is extremely limited. Objects are assumed to be aligned with the background plane. Only meaningful z value increases height of object from plane.

Creating 3D objects seems unnecessarily difficult. When you think about it, we don’t live in a full 3D world. Most of our surfaces are flat and at 90 degrees. There isn’t software that can create actually useful 3D content. Everything seems to be full 3D or nothing.

2D is of course a subset of 3D. Apple offloaded Quartz 2D rendering to the GPU in Tiger, 20 years ago.

Instead of trying to make something look 3D, just make it in 3D.

A logical conclusion to draw from this is why not just have everything in 3D? I’m not saying flythough UI, but an orthographic view that looks very similar to today’s 2D interfaces, but everything is 3D, buttons, text, controls, etc.

… like lighting and shadows or 3D shaders like Liquid Glass. The whole UI is effectively running in a 3D accelerated world where windows are polygons which can be animated, etc. So it’s very much 3D -> 2D -> 3D.

From a design perspective people are either using 3D tools directly, or using something like Photoshop to create 3D effects like lighting, shadows, shading. These must be transformed to 2D bitmaps for the UI framework. But the UI framework itself uses GPU effects…

One of the interesting things about UI is that a lot of the design is 3D, or at least pseudo-3D, then gets forced through a 2D API which then runs on a 3D GPU. 3D -> 2D -> 3D Can we stop pretending and skip the 2D part?