On Tue, May 18, 2021 at 01:27:56AM +0200, Cyril Brulebois wrote: >Simon McVittie <s...@debian.org> (2021-05-17): > >> I also think the beginning of Debian 12 would be a good time to >> reconsider whether the graphical d-i mode is really the best way for >> non-expert users to install Debian. The restricted capabilities of >> udebs make d-i quite a "one hand tied behind your back" environment, >> which was still a necessary evil a few years ago; but now that systems >> with 512M RAM are literally given away with a magazine, perhaps that's >> becoming less necessary than it once was? > >Maybe. > >Until now, I've been happy with maintaining (or at least trying to >maintain) the status quo, which means an installer that works both in >text and graphical modes, which can be preseed “as usual”, etc. It's >definitely not getting many extra fancy features, but it seems to me >it's been working rather reliably for a number of users, so… > >Of course it comes with a price regarding debuggability and you're >unfortunately the one paying here, and I'm sorry for that.
So my own idea for how to proceed here is to make d-i a less restricted environment. IMHO most of the core design of d-i is still very gdoo, but it's a little too limited by its focus on small systems (32MiB on the NSLU2, etc.) that are just not a target any more. We could quite readily improve some of the more difficult areas of today's d-i by adding support for (say) Python 3 instead of sticking to the existing mix of mostly shell with C and perl bits. Even if was decided to recommend that new users use live media for installations, the flexibility of d-i is massively powerful, and we shouldn't give up on it. The ability to support everything from a serial terminal up to a graphical installer on the same media is lovely. But... I'd rather not start a discussion here about the future of d-i, right at the end of a release cycle. Instead, I think we could do with some focused face-to-face sessions at a debconf (or maybe a targeted sprint) to work through the requirements and options properly. -- Steve McIntyre, Cambridge, UK. st...@einval.com “Changing random stuff until your program works is bad coding practice, but if you do it fast enough it’s Machine Learning.” -- https://twitter.com/manisha72617183
