Small Rust fixes

[acfoltzer]

A couple small fixes to improve quality of life for Rustaceans:

First, pass the target name to rustc as a crate-name argument, rather than relying on pragmas within the Rust source files. This reduces the extent to which existing Rust sources must be modified in order to work under Meson.

Second, use the default Ninja behavior for rustc-generated depfiles. This prevents unnecessary rebuilding; see c93ef9b.

[codecov-io]

Codecov Report

Merging #2053 into master will increase coverage by <.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #2053      +/-   ##
==========================================
+ Coverage   64.01%   64.01%   +<.01%     
==========================================
  Files          72       72              
  Lines       17222    17223       +1     
  Branches     3550     3550              
==========================================
+ Hits        11025    11026       +1     
  Misses       5106     5106              
  Partials     1091     1091

Impacted Files	Coverage Δ
mesonbuild/backend/ninjabackend.py	88.24% <ø> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bff37a9...1e4cddc. Read the comment docs.

[jpakkane]

The deps = gcc rules delete the depfile after it has been read into Ninja's database. This conflicts with the fact that rustc puts a target for the depfile itself within the depfile, and so the deleted depfile triggers a rebuild of the whole target.

This seems like a bug in rustc because it is different behaviour from every other compiler. Maybe this could be reported upstream and fixed properly there? Possibly we could report this to Ninja as it is the one deleting the depfile.

[acfoltzer]

I took that commit out of the branch so that the crate-name can be merged before we figure out how to address the depfile. I can chat with the Rust folks, but are you aware of why Ninja deletes the file in this case?

[jpakkane]

It's a performance optimisation. In Windows creating thousands of small files is very slow so Ninja deletes them immediately so they never touch the disk.

[acfoltzer]

Gotcha. I'll see if I can get the Rust folks to change this behavior, or at least add an option to remove the depfile target from the output.