rain boosted

@rain @sirgazil Actually got nerd-sniped by this, and now writing a simple doc generator script (in Chicken, mostly portable) so at least I have something usable. Should be done tomorrow, as of this morning it parses "" and #<<DOC strings, but doesn't identify functions yet.
#scheme

rain boosted

@rain >Where did that prebuilt binary come from?

I mean that's true no matter what isn't it? How do you get an assembler without already having one? How do you go from bare metal to an assembler without using a computer? I actually don't know how and it kind of bothers me.

There's the old joke about a magnetized needle and a steady hand but manually flipping bits on common mediums sounds impossible.

rain boosted

@rain
go back far enough and someone had to write the first compiler tool in assembly

rain boosted

@aeveltstra this is actually true in practice, guix builds rust by building 'mrustc' (written in C++) then building 16 increasing versions of the rust compiler until it reaches rust 1.35

rain boosted

@rain What comes to mind is that an initial compiler is made using a different toolset (like an initial rust compiler written in c). And once that initial compiler exists, it can be used to create a compiler that recreates itself.

is about building a compiler using tools smaller than itself, as opposed to building a compiler using an already built version of itself.

The problem with the second is: Where did that prebuilt binary come from?

the bank is displaying a huge pride rainbow

rain boosted

What are your favorite documents about ?

rain boosted

GNU Mes 0.19 released! Featuring performance; Now bootstrapping
@gnutools GCC in ~30min. Started audit and verify thereof in
#nixos, support in Mes C Lib to build #gnubash and #gnutar.
#bootstrappable @ReproBuilds
-- lists.gnu.org/archive/html/inf

rain boosted
@KitRedgrave @janneke @ky0ko Yep yep. Whatever decade I get to the point where I have enough hubris to write my own language, it'll definitely be bootstrappable from basically metal. I won't even use C, just go straight from machine code to Forth to Scheme to Language.

I'm happy to see the #bootstrappable and #mes people talking about the faraway goal of even taking the kernel out of the picture.

If you asked me what the 4 best documents regarding bootstrapping are i'd say:

* Egg of the Phoenix (Blog post) - canonical.org/~kragen/eotf/

* The Cuniform Tablets of 2015 (Blue-sky academic research) - vpri.org/pdf/tr2015004_cuneifo

* Preventing The Collapse of Civilization (Video) - youtube.com/watch?v=pW-SOdj4Kk

* Coding Machines (scifi story about trusting-trust attack) - teamten.com/lawrence/writings/

If you asked me what the 3 most important documents regarding are i'd say:

* Egg of the Phoenix - canonical.org/~kragen/eotf/
* The Cuniform Tablets of 2015 - vpri.org/pdf/tr2015004_cuneifo
* Preventing The Collapse of Civilization - youtube.com/watch?v=pW-SOdj4Kk

rain boosted

@rain @sirgazil That's my current practice, but I don't have the macro defined yet. I may switch to the hahn egg, since it's most JavaDoc-like, or see if I can get Doxygen to recognize Scheme comments.
wiki.call-cc.org/eggref/4/hahn

Current inline doc situation stinks.
#scheme

rain boosted

I just posted a comment on the orange website at news.ycombinator.com/item?id=2 about #bootstrapping/#bootstrappable stuff. I hope it's not too unfair that I only called out OriansJ and rain1 by name in it. I'm interested in feedback, especially during the editing window, in case I said something dumb or missed a key point I should have, or just wrote sentences that are hard to read or understand, or said something true but ill-advised.

rain boosted

@sirgazil if you want machine extractable comments use strings inside function body and have define register that string as documentation in a table rather than trying to parse the source code. the reason for doing it this way is that it will work cleanly with macros.

Show more
niu.moe

Welcome to your niu world ! We are a cute and loving international community O(≧▽≦)O !