| 13 Jan 2024 |
 samueldr | oh, there's also that libc one, didn't catch on the diff | 23:21:45 |
 raitobezarius | yeah there's a whole machinery to slowly transform your thing into final packages and stdenvs | 23:22:10 |
| raitobezarius | I don't think you can overlay the lib/systems ones though | 23:22:23 |
| raitobezarius | I'm still not sure what to advise as I'm not sure what is your intent, if you don't want to bother nixpkgs with this platform… is your optimal goal to have an overlay somewhere where you can build things with it? | 23:22:58 |
| samueldr | whatever mechanism to achieve my goal (editing Nixpkgs or not) is okay | 23:25:01 |
| samueldr | I want to better understand the non-trivial cross-compilation | 23:25:13 |
| samueldr | a precise goal would be to have an "hello" world C program cross-compiled through as much of stdenv.mkDerivation as feasible | 23:26:37 |
| samueldr | (the GNU one most likely can't) | 23:26:49 |
| 14 Jan 2024 |
| samueldr | oh, right, another thing I don't know what to do about: the quatuplet for config, since the (admiteddly old) SDK and all the few remnants of info only specify a triplet | 00:24:41 |
| raitobezarius | (going to sleep for now but I have some ideas) | 03:52:06 |
| raitobezarius | In reply to @samueldr:matrix.org
oh, right, another thing I don't know what to do about: the quatuplet for config, since the (admiteddly old) SDK and all the few remnants of info only specify a triplet I think you can invent whatever you want and if you feel like it, you can even send it upstream to gnu-config | 17:05:24 |
| raitobezarius | I think it would help a lot if you could share a repo or a branch of nixpkgs so that I can take a look at things | 17:05:42 |
| raitobezarius | first step is that I would wire up the libc properly, create a lib/systems/examples.nix for your platform/target thingie, wire the libc in there, then I would add some inspection patterns for this platform, then I would look into stdenv/cross/default.nix and add the special cc for it | 17:07:27 |
| raitobezarius | then… I'd take it for a run with pkgsCross.<my-config-name>.hello ? | 17:07:38 |
| raitobezarius | * then… I'd take it for a run with pkgsCross.<my-example-name>.hello ? | 17:07:45 |
| samueldr | okay, so papercut 1:
checking target system type... Invalid configuration 'i386-unknown-btron-gnu': Kernel 'btron' not known to work with OS 'gnu'.
| 21:25:49 |
| samueldr | (I'm not done) | 21:26:09 |
| samueldr | so uh, "OS 'gnu'... weird" | 21:26:19 |
| samueldr | mostly to make my headcanon more complete, I try i386-unknown-btron-btron in Nixpkgs | 21:26:59 |
| samueldr | Unknown ABI: btron | 21:27:07 |
| samueldr | yeah, my understanding was that the last bit was the ABI, not the "OS" | 21:27:19 |
| samueldr | is it that gnu config thing having other definitions than what Nixpkgs uses? | 21:27:33 |
| samueldr | * is it that gnu config thing having other terminology than what Nixpkgs uses? | 21:27:43 |
| samueldr | still working out the suggestions you made in a branch, will come back with something | 21:29:34 |
| * samueldr grumbles | 21:30:49 |
| samueldr | https://www.gnu.org/software/autoconf/manual/autoconf-2.68/html_node/System-Type.html#System-Type | 21:30:50 |
| samueldr | no definition of os in here | 21:30:55 |
| raitobezarius | In reply to @samueldr:matrix.org
is it that gnu config thing having other terminology than what Nixpkgs uses? I think the situation is that we have diverged quite a bit from the gnu-config parser, yes | 21:31:54 |
| raitobezarius | There's a PR to put us back on the right rails | 21:31:59 |
| raitobezarius | And John Ericson and Adam Joseph have been sending patches to gnu-config to correct them too | 21:32:11 |
