| 18 Oct 2025 |
matthewcroughan | I mean it might build, but who knows whether it works at runtime with mercurial the same exact way that is expected | 16:06:30 |
dramforever | i checked the code in mercurial and it does runtime checks for feature testing | 16:07:28 |
dramforever | like for example if it's missing test_match it uses match instead | 16:07:42 |
dramforever | oh they even have code that add fb-re2 compat https://phab.mercurial-scm.org/D10284 | 16:11:18 |
dramforever | so mercurial most definitely does not specifically depend on fb-re2 | 16:12:48 |
dramforever | and given that the actual regex works the same under the hood, i really doubt there will be any problems | 16:13:12 |
dramforever | oh they also catch re2 errors and fall back to normal re | 16:13:44 |
matthewcroughan | so they even noticed that the code was bad and then fallback | 16:19:10 |
matthewcroughan | lolwut | 16:19:11 |
matthewcroughan | Bah so close to making a fully switchable system by swapping Nix out for Lix | 16:19:47 |
matthewcroughan | But Lix fails with llvm in ways that nix doesn't | 16:20:01 |
matthewcroughan | FAILED: [code=1] lix/libutil/liblixutil.so.p/processes.cc.o
aarch64-unknown-linux-musl-clang++ -Ilix/libutil/liblixutil.so.p -Ilix/libutil -I../lix/libutil -I. -I.. -flto -fdiagnostics-color=always -D_LIBCPP_HARDENING_MODE=_LIBCPP_HARDENING_MODE_FAST -fpch-instantiate-templates -include-pch lix/libutil/liblixutil.so.p/precompiled-headers.hh.pch -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -Wextra -std=c++23 -O2 -g -DLIX_UR_COMPILER_UWU -fp>
../lix/libutil/processes.cc:249:17: warning: parentheses were disambiguated as redundant parentheses around declaration of variable named 'stdout' [-Wvexing-parse]
249 | std::string stdout;
| ^~~~~~
/nix/store/4w3bhgsi1al9v1jwgk9vc6ch20l7ynba-musl-aarch64-unknown-linux-musl-1.2.5-dev/include/stdio.h:67:16: note: expanded from macro 'stdout'
67 | #define stdout (stdout)
| ^~~~~~~~
../lix/libutil/processes.cc:249:17: note: add a variable name to declare a 'std::string' (aka 'basic_string<char>') initialized with 'stdout'
249 | std::string stdout;
| ^
| varname
/nix/store/4w3bhgsi1al9v1jwgk9vc6ch20l7ynba-musl-aarch64-unknown-linux-musl-1.2.5-dev/include/stdio.h:67:16: note: expanded from macro 'stdout'
67 | #define stdout (stdout)
| ^
../lix/libutil/processes.cc:249:5: note: add enclosing parentheses to perform a function-style cast
249 | std::string stdout;
| ^
| ( )
../lix/libutil/processes.cc:249:17: note: remove parentheses to silence this warning
249 | std::string stdout;
| ^
/nix/store/4w3bhgsi1al9v1jwgk9vc6ch20l7ynba-musl-aarch64-unknown-linux-musl-1.2.5-dev/include/stdio.h:67:16: note: expanded from macro 'stdout'
67 | #define stdout (stdout)
| ^
../lix/libutil/processes.cc:384:25: error: use of undeclared identifier 'SYS_capset'
384 | if (syscall(SYS_capset, &hdr, data)) {
| ^~~~~~~~~~
1 warning and 1 error generated
| 16:20:18 |
dramforever | not really bad, more that fb-re2 didn't do bytes | 16:20:32 |
dramforever | but i guess in a way | 16:20:39 |
dramforever | that means it's python 2 brained | 16:20:49 |
dramforever | okay this is really funny | 16:21:23 |
matthewcroughan | dramforever: hahahaha | 22:50:01 |
matthewcroughan | security.enableWrappers = false; makse you unable to login 🙈 | 22:50:10 |
matthewcroughan | but is also completely necessary for native musl/llvm compilation | 22:50:30 |
matthewcroughan | * but is also completely necessary for native musl/llvm compilation (due to pkgsStatic being broken) | 22:58:38 |
matthewcroughan | * but is also completely necessary for native musl/llvm compilation (due to pkgsStatic being broken on it) | 22:58:45 |
| 19 Oct 2025 |
dramforever | who needs to log in anywa | 08:10:51 |
dramforever | * who needs to log in anyway | 08:10:53 |
Grimmauld (any/all) | You wanna try https://github.com/thkukuk/pwaccess maybe, that would make pam/shadow work without suid | 08:45:42 |
Grimmauld (any/all) | it was on my list of things to try, but if you do get it working then definitely send me a snippet! | 08:46:47 |
Grimmauld (any/all) | actually i may try cooking something up today for a suid-less login... We'll see how successful i am | 08:57:14 |
matthewcroughan | Should be able to make a VM test for it right? | 10:09:27 |
Grimmauld (any/all) | yep | 10:09:39 |
matthewcroughan | Sadly musl vm test isn't possible atm because some things break | 10:10:51 |
matthewcroughan | error: builder for '/nix/store/34dsfyf2hxmdkc4aahwvwrn22m45ki07-perl5.40.0-Alien-Build-2.80-aarch64-unknown-linux-musl.drv' failed to produce output path for output 'devdoc'
| 10:13:28 |