Closed
Description
Hi, I'm setting up a checkout of the Rust repository on a Windows machine. I would like to run x.py setup
to generate a config.toml
file with the library
profile. x.py
aborts due to ninja not being installed, even though this does not appear to be necessary for the library profile because x.py
fetches pre-built copies of LLVM.
To work around this I had to copy the example config and manually modify the ninja setting to false. Running x.py setup
again returns an error since the config file already exists, although it does suggest setting a top-level profile
key.
This is a speedbump on the road to contributing.
Activity
x.py setup
#84940jyn514 commentedon May 19, 2021
Note: to reproduce this, run
x setup library
(the error only occurs after the prompt for which profile).kornelski commentedon Jun 1, 2021
I've ran into this too. It'd be nice if setup could put ninja=false in config.toml automatically. Otherwise it's a funny situation where it requires config.toml in order to create the config.toml.
kornelski commentedon Jun 1, 2021
The error message doesn't say in which section to put
ninja=false
in! Justninja=false
alone in the file causes a TOML parse error. I've tried putting it in a[build]
section, but it didn't like it either.jyn514 commentedon Jun 1, 2021
@kornelski it goes in the llvm section
ninja = false
goes under[llvm]
#85988Rollup merge of rust-lang#85988 - jyn514:ninja-error, r=joshtriplett
Rollup merge of rust-lang#84940 - jyn514:ninja, r=Mark-Simulacrum