Cargo pre-planning: project knowledge

[nrc]

• std-aware/Xargo
• sysroot/extern crate stuff

cc @japaric, @ehuss , @alexcrichton , @joshtriplett

[nrc]

ping @japaric, @ehuss , @alexcrichton , @joshtriplett

Could start to develop the design space/layout the options and pros/cons here please?

[alexcrichton]

Sorry I haven't had a lot of time to dig into this, I was planning on revisiting/posting to this after the 2018 edition is out

[alexcrichton]

Ok I've got some time, so here's some thoughts!

I think the biggest impact thing we can do here is to basically "fix" Cargo's handling of the standard library. This includes things like:

• I'm compiling for MinGW on Linux, I have to remember rustup target add
• I'm compiling for WebAssembly, but I'd like to test out the upcoming atomics feature, I have to remember to use xargo or cargo-xbuild
• I'm compiling a target where there are no official binaries, I have to remember to use xargo or cargo-xbuild
• I'm experimenting with various options like instrumentation, miri, profiling, etc, and I need to remember to recompile the standard libary with various flags
• I'd like to express some degree of configuration about the standard library. For example I don't want backtraces enabled or maybe I want panics to immediately abort with no diagnostics for code size. The list of possibilities here is quite long!

I'm sure there's other entries for this list as well. I think that we can definitely collect enough use cases to make a compelling argument that this should be prioritized. The main con and difficulty will be designing this in a way we're all comfortable stabilizing. There's been some great groundwork historically and we're almost there, I think we just need to give it another final push to get over the design hump before we can make progress.