Janneke Nieuwenhuizen wrote:
Thanks...and hmm, isn't that weird? And what she says also puzzles me:18:46 So, another technique that you could try to use instead is kind of handwriting your own teeny compiler and then using that as a bootstrap to generate larger and larger compilers until you have a final compiler with a pure lineage that has been written entirely by yourself that you know you can trust. This ("that you could try") doesn't sound to me as if Lauriewired is actually aware that the Mes Full Source bootstrap is not just a theoretical possibility, or a pretty good hack/workaround like DDC, but a viable, actively developed, in-production solution to this terribly important problem ("original sin!") "that no one can fix"?
Mes Full Source bootstrap is a great achievement and I agree it should be explicitly mentioned there, but the issue of the "original sin" is a bit bigger in my opinion. My interpretation of "handwriting your own teeny compiler" is to create some kind of dedicated hardware / stack machines like https://en.wikipedia.org/wiki/Harris_RTX_2000 that could first program the x86 coreboot, RISCV sbi/uboot, ARM bootROM/uboot, etc so that the live-bootstrap could be "safely" kick off.
https://www.youtube.com/watch?v=KrksBdWcZgQ
There are > 3K comments and none of the top few hundred "got this". Rather, the recurring remark seem to be something like: This is terrible and someone should fix this/we should start computing anew from the start/computing is cursed, let's give up.
Imagine a hypothetical post-apocalyptic scenario where all binary software is gone (including low-level firmware) so how can we bootstrap the world having all the electronics and docs/source-code in the paper format? I don't think we have an answer for that and for sure not the optimal one, so the situation is indeed quite terrible and seems to be cursed by design, but instead of giving up lets try to challenge and fix it properly.
https://libresilicon.com/ , https://libre-soc.org/ ... Cheers! Martin
