nix: use meson #7774
nix: use meson #7774
Conversation
|
I wonder why though, didn't we explicitly move from Meson to CMake? And I'll make a guess here that only one works here at any given time, presumably CMake. |
|
Oh I guess CMake building of Nix package is being phased out (#7778). If it solves issues like this #7508 (comment), I have no complaints. |
|
There is a possibility, but the main reason we've moved back to meson is due to the faster builds (almost twice as fast). |
|
Yes, as weird is at is, meson builds twice as fast. |
|
sounds wrong and doesn't seem to be the case on my end. CMake builds a bit faster. Have you tried:
|
|
@fufexan ? |
|
2 hours and a bad graph later, I'm back with some rough results. Note that these builds were run on a Ryzen 5800U which is already better than GitHub's CI, but worse than many setups out there.
Looks like we'll go back to using CMake. |
No longer using ninja since CMake seems to build faster with the Make backend (see #7774 (comment)). Using mold for faster linking. Using a stdenvAdapter for keeping debug info if `debug` is true.
No longer using ninja since CMake seems to build faster with the Make backend (see #7774 (comment)). Using mold for faster linking. Using a stdenvAdapter for keeping debug info if `debug` is true.
No longer using ninja since CMake seems to build faster with the Make backend (see #7774 (comment)). Using mold for faster linking. Using a stdenvAdapter for keeping debug info if `debug` is true.
No longer using ninja since CMake seems to build faster with the Make backend (see #7774 (comment)). Using mold for faster linking. Using a stdenvAdapter for keeping debug info if `debug` is true.
CC @fufexan