| 12 Dec 2024 |
 Tristan Ross | I might look into it after the LLVM libc PR is merged | 18:32:57 |
 truby | I don't think it should be too hard; I think you can just wrap it in exactly the same way as clang as all the driver options should be the same. We didn't support -B last time I tried to look at it, if that's still the case I'm happy to add that from the flang side | 18:34:10 |
| Tristan Ross | Maybe, we'll have to see. | 18:35:19 |
| truby | I'm not sure how many of the hardening flags we have either; or how many of them make sense for Fortran. | 18:35:43 |
 Artturin | truby: https://github.com/search?q=lang%3Anix%20%2Fflang%2F&type=code | 18:35:56 |
| Tristan Ross | Also, my CPU model PR I might close and redo completely. I have better ideas as to handling it. | 18:36:40 |
| truby | as another unrelated question; do you know if there's a better solution for having a dev environment for working on LLVM than just using fhsEnv? So that the clang you build in the dev shell can actually find libc etc? Atm I'm just shoving everything in a fhsEnv but it feels kinda hacky. | 18:37:30 |
| Tristan Ross | Check the wiki | 18:37:49 |
| Tristan Ross | I've done nix develop nixpkgs#llvmPackages.clang | 18:38:16 |
| Tristan Ross | And that's worked before. | 18:38:22 |
| truby | I tried what was on the wiki and it didn't work for some reason, but now that I think of it that might have been flang not supporting -B... :) | 18:38:52 |
| truby | guess that's on me to fix if it's still true... | 18:39:50 |
| truby | I think most of these are referring to "classic" flang, which is a completely different compiler with the same name. It's not at all confusing and never causes any issues for us. 😃 | 18:40:41 |
 p14 | In reply to @rosscomputerguy:matrix.org
And that's worked before. That's fine for building clang, but if you subsequently try to use the built clang, it won't work well because it's not wrapped. | 19:22:21 |
| p14 | I'm not sure I've seen nixpkgs have a good answer to this. I think what's needed is a way of dumping the flags needed somewhere (e.g. to find the C/C++ stdlib) that the development compiler can pick up. My method is to use NIX_DEBUG on some other compiler and dump them into a clang{,++}.cfg next to my development compiler. | 19:23:28 |
| p14 | Which is a little fragile since there aren't GC roots for those references, but it seems to work well enough for my purposes as a development environment. | 19:23:56 |
 Philip Taron (UTC-8) | I filled it in. Did you see the draft I posted earlier? | 19:45:57 |
 Randy Eckenrode | I need to add my availability, though it’s probably a bit limited. | 19:46:53 |
| Tristan Ross | What draft? | 19:55:09 |
| Philip Taron (UTC-8) | Hold on, rather than linking the Matrix message I'll just make a pad. | 20:05:31 |
| Tristan Ross | Alright | 20:05:45 |
| Philip Taron (UTC-8) | https://pad.lassul.us/gYaWH4LuQGeED_BUSGInQA | 20:25:19 |
| Tristan Ross | Looks good | 20:26:02 |
| truby | Tristan Ross: da67096a3b9bf56a91d16901293e51ba5b49a27e is my nixpkgs rev. Still hitting the gettext failure | 20:29:04 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/commit/da67096a3b9bf56a91d16901293e51ba5b49a27e | 20:29:36 |
| Artturin | the bump is not yet in master https://nixpk.gs/pr-tracker.html?pr=358254 | 20:29:43 |
| truby | hm | 20:29:44 |
| Artturin | while the removal is | 20:29:45 |
| Artturin | was submitted agaisnt master | 20:29:50 |
| truby | I'm probably misunderstanding the flake file format | 20:29:56 |
