@ljs @ohjames AFAICT #Android for the most part doesn't do much of an #init system and merely uses #toybox to make it's #UI able to run and completely abstract the underlying system.
Plus yeeting the init
system - even if it's just a single /etc/init
file will brick the system and necessitates reflashingbabnew ROM.
@gettie @chesheer Sadly not even #AdelieLinux, which @ActionRetro uses all the time...
I am working on an #i486 distro ( @OS1337) but for #i386 the support in terms of Linux ended with Versions 3.4.99 LTS & 3.6.9 respectably, and my #userland & #toolchain (#toybox & #musl-cross) doesn't support that at all, and I'd propably have more success convincing @landley to join #OS1337 than to support i386 even if I could pay him for that!
@FandaSin @cleverboi @BrodieOnLinux @neal I think #i486 is a good baseline because that's the furthest some systems can go!
@cleverboi @FandaSin @BrodieOnLinux @neal as for #i386 I understood why cuz it was more and more painful m, but the problem with #i486 is that in several #embedded and #industrial setups there are still newly deloyed systems based off it.
I.e. #Vortex86 #SoC's cuz #MSDOS and shit still gets used in #industrial equipment.
Linux stopped supporting i386 with versions 3.4.99 (longterm)
& 3.6.9
respectably.
i386
where none of the toolchain (#musl) and utilities (#toybox) supoort it, i486
is still supported there.And I really want to continue developing a minimalist "rescue" distro that can handle such legacy hardware because it may be the only option to ddrescue
stuff from certain systems or to properly & reproduceably backup & restore them!
@mrmasterkeyboard @landley My goal with @OS1337 is to go beyond the #mkroot functionality of #toybox and allow for something that can be as brutally simple (as like a BOOT-ROM) or elaborate as desired, with the specific focus on being useable in 80x25 MDA.
@landley @mrmasterkeyboard nodds in agreement
Granted the #IETF also accepts proprietary standards like #HSRP which resulted in #CARP being developed.
@landley @mrmasterkeyboard yeah.
Personally, if "producing paper" wasn't undesireable you'd propably consider just taking the subsets of #POSIX and #LSB you're implementing and standardizing that as a sorf-of "#TINYNIX" (MININIX may be confused with Minix, MICRONIX and NANONIX already exist) standard.
I just tend to ask what are the comfort features and tools that I use and want if I have to use "my little distro" for my daily job as linux sysadmin.
@mrmasterkeyboard @landley glad I could get you guys connected then...
I'm convinced #toybox is a good #userland to get a basic OS up and running and on it's feet. If it's good enough for #Android then it's certainly gonna be good enough for #embedded stuff as well...
I don't leverage that capability with #OS1337 (at least for now) but that's because I treat it more akin to how #VxWorks and #WindRiverLinux is being developed and deployed...
@mrmasterkeyboard I mean, you're way deeper into the weeds than I am, and I guess you're competent enough to find any issues with #toybox and let @landley know about them.
@mrmasterkeyboard If that #kernel is mostly #POSIX-compatible it should be possible to port #toybox over there (alongside any other software one may want to get running on @OS1337), but that's as far as I know.
Not shure if @landley has the time and spoons beyond testing toybox against #Linux as he does aim to make it a better alternative to #BusyBox & #GNUtils.
To me @OS1337 is just an attempt to a minimalist #Linux distro because I want some #reproduceable & #auditable #firmware for various other projects, and both @yoctoproject and #RaspberryPiOS 'lite' seem rather excessive to me.
I just think that it can have serious benefits being less distracting and allowing me (and others) to just use basically any hardware to get work done...
@lmemsm basically the Idea behind it is to be a brutally simple #toybox + #musl / #linux distro that grew out of the necessity for me to actually think about #firmware for some projects.
Basically I want something that is so simple and auditable that it's practical to make it pass any #verification demands for #SecureTerminal|s in #CriticalInfrastructure and #Communications.
Not to mention a #GNUfree - #Linux distro is the way to go if I want that thing to not get bricked constantly by minor #GlibC-changes...
I hope that answers your question...
@davidgerard problem is that I can't rely on #python being available and being allowed to install it!
@wyatt Also glancing from the #PC98 architecture and specific quirks that #Linux accounts for in menuconfig
it's most likely not gonna be enough to "just boot the #i486 version of @OS1337"...
https://github.com/OS-1337/OS1337/blob/main/OS1337-core-prerelease.img
As for the specifics of the #PC9821+ and their detailed hardware I'm clueless beyond the Wikipedia Article and can only Assume anything before a (RAM-upgraded) #9821Ap & #9801BA is off the table and a #9821Af or better is desireable as the 14,6MB RAM limit will really become a problem quickly, as I'm pretty shure it's impossible to get linux-6.6.6
run on anything below 8MB at all and most recommendations hint that 16MB is more often than not a practical limit (tho that may be becaise 10/12/14 and other "asymetric" configurations were already avoided back then)...
On the flipside I did manage to install #WindowsXP on machines with 32MB RAM and actually get them to display a desktop (abeit with seconds per frame instead of frames per second) so this is more likely a question of chipping down things.
dd
something on a #BlueSCSI or similar...@kobilacroix nah.
You'd propably find him and Howard doing some weird shit, like an #OpenBSD / #toybox distro or a #DragonflyBSD cluster...