[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Tinycc-devel] [OT] Re: linux-2.4 build scripts for tinycc uploaded
From: |
gz8cx4 |
Subject: |
Re: [Tinycc-devel] [OT] Re: linux-2.4 build scripts for tinycc uploaded - principle issues with x86 |
Date: |
Tue, 26 Nov 2024 10:47:06 +0100 |
Hello Michael,
On Tue, Nov 26, 2024 at 12:30:22AM +0000, Michael Ackermann via Tinycc-devel
wrote:
> The plan is to converge and fully re-integrate a complete
> i486-tcc-linux-musl.iso distribution with bootstrappable.org
This makes an impression that you are trying to solve the bootstrapping
problem, which is, in brief, to avoid trust in any pre-made binaries.
This problem has been solved. (well, actually by VSOBFS;
not-solvable [sic] by bootsrappable.org, which insists on relaying on
a binary seed, occasionally pretending that its hex-codes is a source :)
If you care about bootstrapping, you can safely base your distribution's
bootstrapping on a well-known-source-true Linux image with a tinycc
compiler.
That is, an image pro-v-ably (in contrast to pro-b-ably) corresponding
to its source code, without assuming any certain toolchain, buildable on
virtually any platform where you can run tinycc (another ANSI-C compiler
of your choice would work too).
Such an image is exactly what VSOBFS offers. You do not have to reimplement
this part, which is hard and tricky, inside your prospective distribution.
On Tue, Nov 26, 2024 at 08:29:07AM +0000, Michael Ackermann via Tinycc-devel
wrote:
> With bootstrapping, particular circular dependencies must be avoided, such as
> relying upon a gigantic dependency graph of software at a stage of
> bootstrapping
> such software could not be available yet. Think about it carefully.
I did think about it carefully, because this has been my interest for
years. That is how VSOBFS became possible.
Having said that, a platform bootstrapping by itself does not concern
tinycc directly, even though tinycc has been an invaluable tool for VSOBFS.
I suggest we move the corresponding part of our discussion, if any,
off the list.
> I haven't had time yet to
> throw my i486-tcc-musl.squashfs at linux-2.4.37.11-tcc kernel,
Ah I see. This is where you can expect problems, unfortunately.
Cheers,
/tccm