IRC channel logs

<aggi>attila_lendvai: it is two known use cases assembly language is critical to bootstrapping

<aggi>1) the early stage0/M2/macro-assembly to spawn a C-compiler with

<aggi>2) the necessity of any assembler to process GRUB/syslinux/kernel-bootcode.S pieces to make any system bootable (on x86_32|16)

<aggi>those two use-cases are not directly related, hence different assemblers/sytax is involved with either

<aggi>however, it might be an interesting peculiarity of current live-bootstrap

<aggi>components such as bootloader/GRUB/syslinux and kernel which are required first during system boot

<aggi>ironically those components are bootstrapped last with live-bootstrap because GNU binutils/gas is rather complicated

<aggi>hence binutils cannot be bootstrapped early in live-bootstrap as it should be similar to the stage0/M2 macro-assembly

<aggi>bin86/as86 or yasm seem a little less troublesome, but can't break the circular dependency either

<aggi>yasm at least supports gas-syntax, bin86/as86 doesn't - so whichever assembler was chosen, syntax and mnemonics must match

<aggi>there's a chance syslinux might pass with yasm to choke isolinux.bin (which is the preferred target with tiny-bootstrap to cdrecord the bootstrap output early)

<aggi>no clue about RISC-V and kernel/bootloader related pieces; in the end there seems no way around binutils/gas to support multiple ARCH

<attila_lendvai>aggi, interesting, thanks!

<attila_lendvai>my impression is that there are two bird's eye view use-cases here: one of them is the trusting trust story, and the other is the CollapseOS story. the former allows capturing compiled output if it's small and auditable, e.g. for an assembler. the latter doesn't, because it wants to be able to redo the bootstrap with a "reasonable" effort when we need to bootstrap civilization itself... does

<attila_lendvai>this make sense?

<matrix_bridge><Jeremiah Orians> Very much so.

<aggi>attila_lendvai: it's noteworthy too even nowadays when supply chains remained mostly intact a sufficiently complete system-bootstrap for RISC-V was not yet fully established

<aggi>meanwhile nvidia reports a trillion on their accounts

<attila_lendvai>aggi, that implies that the bootstrap is fully RISC-V, right? i.e. no other arch is allowed to be involved. i guess so, because that (cross-compilation) is what enables a functional RISC-V linux today...

<matrix_bridge><Jeremiah Orians> The minimal RISC-V firmware is a good bit larger than should be required to bootstrap (but I can see why they adopted the standard UEFI)

<aggi>attila_lendvai: not necessarily; say a decent bootstrap avoids circular dependencies (compile-time)

<aggi>such as, not relying upon a POSIX development host before that got fully bootstrapped itself

<aggi>nowadays it's been merely coincidence you rather avoid any typical x86 *nix and cross-compilation from there

<aggi>RISC-V by itself did not yet significantly improve upon x86, even when it's free/open/royalty-free

<aggi>if anytime a CollapseOS type system was desired, it's probably some calculator type system (TI-84 or much simpler even)

<aggi>and i doubt at least this could be manufactured nowadays in compliance with various acceptance criteria of bootstrapping

<roconnor>what would signficant improvement upon x86 entail?

<attila_lendvai>not sure i totally follow... with that in mind: my friend made a Forth-like (https://github.com/nagydani/seedling/) that he intentionally designed so that he (!) must be able to bootstrap the Z80 binary with pen and paper. and he did it... :)

<aggi>ok, i mean, if X86 was a reference platform which got fully bootstrapped down to hardware/firmware and appeased free/open developers

<aggi>it would be less critical to cross-compile for any ARCH from there

<aggi>to improve upon X86_32|16 (which reached obsolescence meanwhile), requires a separate discussion and systematic approach

<aggi>and, i doubt it's feasible even nowadays when global supply chains remained intact

<aggi>best we've got is ao486 FPGA deployment, and i'm not in the mood to fiddle with Altera Quartus and similar

<aggi>for that matter i doubt we've had such a workstation to develop and deploy any Z80 SoC at least during a CollapseOS scenario

<aggi>and, i tend to think it won't be some solar eclipse doomsday type collapse, but a civilizational decline over several decades

<aggi>barely noticable, each day a little more science and knowledge is lost

<aggi>meanwhile nvidia is bragging over their trillion budget and high-res space telescopes broadcast 4K bitmaps from the skies

<aggi>but if you look behind the glamorous facade of it, it's far less impressive than apollo era or forth type systems (which predate my birth)

<attila_lendvai>yeah. reading some very old IT papers makes one wonder whether we are in another dark age...

IRC channel logs

2026-08-30.log