l

> How did you create the current flake.nix? Because it seems like there is a lot of repetition in there Wrote it over time. It's not autogenerated or something.

On git commit e117efa74495867b07400924dc42dade9cb84381 ``` % git diff % git diff diff --git a/flake.nix b/flake.nix index bd532030..8858deac 100644 --- a/flake.nix +++ b/flake.nix @@ -358,6 +358,8 @@ formatter = pkgs.nixpkgs-fmt; packages...

Tried to compile on GitHub Actions: ``` ... building '/nix/store/97g442r55nxy8hy3c48j47a1gn81n16f-stdenv-darwin.drv'... building '/nix/store/9cig52gmhg0h6yxlxp9ckshwm6h240l4-compiler-rt-libc-x86_64-darwin-16.0.6.drv'... building '/nix/store/chd1lijgrrpg6zp78mj2ccc0gc2ic0s8-x86_64-darwin-clang-wrapper-16.0.6.drv'... building '/nix/store/ihd2xpqlyxmbk8ck3my1xiyxd34b5jps-deltachat-deps-1.149.0.drv'... building '/nix/store/j8xqxzlw1wswxiwbqqvmjghryimnz0dd-deltachat-1.149.0.drv'... error: builder for '/nix/store/j8xqxzlw1wswxiwbqqvmjghryimnz0dd-deltachat-1.149.0.drv' failed with exit code 101; last 25 log...

Draft PR is at https://github.com/deltachat/deltachat-core-rust/pull/5326

I opened two stacked PRs: 1. https://github.com/chatmail/core/pull/7311 (cleanup) 2. https://github.com/chatmail/core/pull/7307 (fix cross-compilation to x86-64 macOS) `nix build .#deltachat-rpc-server-aarch64-darwin` still fails because of upstream issue https://github.com/NixOS/nixpkgs/issues/413910

@jgm > As I've noted on the linked issue, this will very likely have unexpected effects in various places, since all of the parsers have been developed under the assumption...

@jgm > For example, does the template engine assume that CRs have been stripped from templates? Will there be changes if you do a header-include? Just checked, templating engine does...

Grep for `toText` and `toString` (which calls `toText`) does not return that many results, none of them looks like a place to insert `crFilter` for me.

@jgm > See the docs for System.IO.NewlineMode > ... > So, I think part of the motivation for stripping CRs was to ensure that our strings conform to this invariant....

@jgm > Note that System.IO only supports CR and CRLF newlines. I assume it is a typo and read it as "LF and CRLF". > No, the idea was that...