> I spent time explaining what I meant by each of these, relating them to the mathematical definition of the input and how the output matched that.
You only mentioned fonts and what you mentioned was only true about vector fonts without any hinting information (which is what allows fonts to be displayed as the designer intended on fixed grid output like a pixel grid). It is not true for vector fonts with hinting information nor about bitmap fonts.
And what your issue really was was about the binary-only representation, which basically means that what you wanted is subpixel accuracy - ie. introducing antialiasing. This is something that is actually subjective as many people (me included, as i already wrote) prefer the binary approach.
And indeed while you are right about the very specific case of how vector fonts are displayed without enough precision, even when a font renderer does allow for subpixel precision many people prefer to alter the output from what the designer would do for -in their opinion- better results: this is one of the oldest differences between people preferring Mac OS X's font rendering (no alterations) over Windows rendering (alterations to make fonts look sharper). Many font rendering setups you'll find on Linux provide control over this too and while you can objectively say which setup would be more "correct", what exactly looks better is down to the user.
Anyway, this was all about a specific case of font rendering and nothing else.
> If you wanted to, you could probably write a function to quantise how the output for different renderers diverged from the input. CDE would not do well compared to modern systems like macOS.
CDE is not a renderer, it uses Motif and Xlib (and perhaps xcb in places) for its graphics.
> Do you think CDE is making an intentional stylistic choice to render text in this way? Or do you think it doesn't have the functionality to render it with more fidelity to the font definition?
CDE uses Motif and Xlib, Motif does have the ability to render antialiased text via Xft which uses FreeType which is pretty much what most other toolkits use to perform font rasterization.
You only mentioned fonts and what you mentioned was only true about vector fonts without any hinting information (which is what allows fonts to be displayed as the designer intended on fixed grid output like a pixel grid). It is not true for vector fonts with hinting information nor about bitmap fonts.
And what your issue really was was about the binary-only representation, which basically means that what you wanted is subpixel accuracy - ie. introducing antialiasing. This is something that is actually subjective as many people (me included, as i already wrote) prefer the binary approach.
And indeed while you are right about the very specific case of how vector fonts are displayed without enough precision, even when a font renderer does allow for subpixel precision many people prefer to alter the output from what the designer would do for -in their opinion- better results: this is one of the oldest differences between people preferring Mac OS X's font rendering (no alterations) over Windows rendering (alterations to make fonts look sharper). Many font rendering setups you'll find on Linux provide control over this too and while you can objectively say which setup would be more "correct", what exactly looks better is down to the user.
Anyway, this was all about a specific case of font rendering and nothing else.
> If you wanted to, you could probably write a function to quantise how the output for different renderers diverged from the input. CDE would not do well compared to modern systems like macOS.
CDE is not a renderer, it uses Motif and Xlib (and perhaps xcb in places) for its graphics.
> Do you think CDE is making an intentional stylistic choice to render text in this way? Or do you think it doesn't have the functionality to render it with more fidelity to the font definition?
CDE uses Motif and Xlib, Motif does have the ability to render antialiased text via Xft which uses FreeType which is pretty much what most other toolkits use to perform font rasterization.