Hi Alexander,
Thanks a lot for your feedback!
Regarding CUPS: I'm aware that classic CUPS drivers/PPDs are supposed to be replaced by another architecture eventually, though to be honest I don't know much about it yet. Since the actual protocol and compression logic in this driver is a small, self-contained Rust codebase, wrapping it into a Printer Application later on shouldn't be too painful once that becomes the standard on Linux distros.
Regarding ESC/P-R vs. direct per-channel control (and your second comment):
My main goal for this project was quite simple: I wanted to get (at least) the same print quality on Linux that I get on Windows.
Originally, there were two ways to do that:
- implement a classic per-channel Gutenprint driver, or
- see how the official Windows driver does it.
In fact, I started with Gutenprint. But that is a whole different story — you have to make hundreds of test prints and a lot of measurements. And you may need more experience than I have.
In the process, it turned out that with newer Epson printers, part of the magic happens inside the printer's firmware, which makes the driver much easier to implement. And that's exactly what the Windows and Mac drivers use. Instead of hundreds of test prints, I only had to make 20 or so, and I can directly use the official ICC profiles from third-party paper manufacturers.
That said, I completely agree with your point: from a RIP developer's perspective, ESC/P-R locks you into Epson's RGB firmware pipeline, whereas direct per-channel control gives you maximum freedom and can squeeze even more out of the printer. Building such a driver is much more complex than my current driver, though. If you have ideas or know-how on how to do that, feel free to dump them into GitHub issues or discussions — I'm open to integrating it!
My main fear is that such a per-channel driver path needs a lot of profiling work. And if someone uses that functionality without the right profiles, the output would be ugly, right?
And yes, fair point on the article — the AI-written documentation definitely promoted the ESC/P-R solution a bit too much. I've just updated `docs/why-not-gutenprint.md` on GitHub so that the limitations and drawbacks of ESC/P-R are explained.
Best Regards,
René