I'm sure the idea here was a physical quine, although since it only contains 2.5% of its own G-code it's not really a quine, any more than a "Hello World" program is a quine since the string "Hello World" is in the program text. It would be trivial to generate something like this depending on which part of the G-code you pick.
Many G-code dialects contain subprogram calls, loops, and conditionals that seem like they'd permit a physical quine. The standard RepRap firmware breaks the rules by taking input:
Is anyone else confused by thier cookie consent banner? The switches start out gray and become black when toggled. which position means consent? It feels intentionally misleading.
Rant: That type of slider-switch is an inferior usurper of the classic tickbox, that rode in on a wave of touch-screen-ification. Oh, it can be done well, but it's too easy to do it badly.
In this case of useless colors, I think the implicit rule/clue is "move the dot-nub towards what you want", so moving right is indicating you like the text-line, and moving left indicates you don't.
> Manual contains only 2.5 percent of its own G-code in its first version. That low figure is part of the point. Current FFF 3D printing resolution and text scale place limits on how much code can fit onto the object while also describing the volume of the object itself. A fully self-contained version would enter an endless loop, since every printed mark would add more data to be described.
The fact that quines exist means that it must be possible to print a fully self-describing book of this sort, though it's possible that you'd require a more expressive language.
There's probably a way to make an xz file that decompresses to g-code for itself. But plain g-code is not powerful enough because it's just a list of movements due the tool head to perform - no computation.
`M98` allows for essentially function calls (nested or not.)
If you had a part of a machine that could save state (say.. turning on a coolant pump..) I wonder how much more of a turing machine you could wrastle into it.
(or you could just cheat and use one of the hundreds of gcode variants that have computational stuff stapled into them like the Fanuc equivalents, but that's sorta dishonest for the exercise)
And weirdly, neither does Klipper, despite having all the resources in the world to do so. Just not a priority since slicers don't produce code like that, and 99.999% of Klipper's job is to eat whatever a slicer sends it.
So suppose I attached an extruder to a Haas mill or something...
It's fascinating how revered people are who talk in metaphor and implication like this about relatively simple things, when far more complicated things are happening all the time inside their devices.
G-code by itself only describes the series of motor movements over time, though I think it would be practically feasible to have a book contain an executable script, which would generate the same g-code as the one used to print the book. It would be really interesting to see how large the resulting book is.
"This prompt, when fed to a sufficiently capable LLM, will generate the G-code to produce a 3D printed model of a book containing this quote, verbatim, in raised letters."
15 comments:
I'm sure the idea here was a physical quine, although since it only contains 2.5% of its own G-code it's not really a quine, any more than a "Hello World" program is a quine since the string "Hello World" is in the program text. It would be trivial to generate something like this depending on which part of the G-code you pick.
Many G-code dialects contain subprogram calls, loops, and conditionals that seem like they'd permit a physical quine. The standard RepRap firmware breaks the rules by taking input:
https://reprap.org/wiki/G-code#M98:_Call_Macro.2FSubprogram
but Fanuc and other dialects allow M97 to call a subroutine chained to the end of a program.
Is anyone else confused by thier cookie consent banner? The switches start out gray and become black when toggled. which position means consent? It feels intentionally misleading.
Rant: That type of slider-switch is an inferior usurper of the classic tickbox, that rode in on a wave of touch-screen-ification. Oh, it can be done well, but it's too easy to do it badly.
In this case of useless colors, I think the implicit rule/clue is "move the dot-nub towards what you want", so moving right is indicating you like the text-line, and moving left indicates you don't.
This is how you get statutory warning labels and nutrtion information labels on packages.
Because these folks always want to do the least legal thing allowed by law.
Cool idea for printing braille. I bet I could vibe code a program to convert .epub files to braille then to printable copies.
> Manual contains only 2.5 percent of its own G-code in its first version. That low figure is part of the point. Current FFF 3D printing resolution and text scale place limits on how much code can fit onto the object while also describing the volume of the object itself. A fully self-contained version would enter an endless loop, since every printed mark would add more data to be described.
The fact that quines exist means that it must be possible to print a fully self-describing book of this sort, though it's possible that you'd require a more expressive language.
There's probably a way to make an xz file that decompresses to g-code for itself. But plain g-code is not powerful enough because it's just a list of movements due the tool head to perform - no computation.
`M98` allows for essentially function calls (nested or not.)
If you had a part of a machine that could save state (say.. turning on a coolant pump..) I wonder how much more of a turing machine you could wrastle into it.
(or you could just cheat and use one of the hundreds of gcode variants that have computational stuff stapled into them like the Fanuc equivalents, but that's sorta dishonest for the exercise)
_If_ one is using a firmware which supports that.
Grbl, which many of the 3D printer firmwares are based on, does not (and no variables, or loops, or branching).
And weirdly, neither does Klipper, despite having all the resources in the world to do so. Just not a priority since slicers don't produce code like that, and 99.999% of Klipper's job is to eat whatever a slicer sends it.
So suppose I attached an extruder to a Haas mill or something...
It's fascinating how revered people are who talk in metaphor and implication like this about relatively simple things, when far more complicated things are happening all the time inside their devices.
G-code by itself only describes the series of motor movements over time, though I think it would be practically feasible to have a book contain an executable script, which would generate the same g-code as the one used to print the book. It would be really interesting to see how large the resulting book is.
"This prompt, when fed to a sufficiently capable LLM, will generate the G-code to produce a 3D printed model of a book containing this quote, verbatim, in raised letters."
Compression could be interesting