Replace rouille with warp

[Saruniks]

Part of porting changes from cachepot #1620, paritytech/cachepot#131

[Xuanwo]

Wow, great work! Thanks in advance; Please ping me while you think this PR is ready.

By the way, could you add @montekki as a co-author on this PR? I think it would be great to give them recognition.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 7.54717% with 49 lines in your changes missing coverage. Please review.

Project coverage is 41.09%. Comparing base (0cc0c62) to head (134e1a2).
Report is 71 commits behind head on main.

Files	Patch %	Lines
src/dist/http.rs	0.00%	26 Missing ⚠️
tests/system.rs	23.52%	13 Missing ⚠️
src/dist/client_auth.rs	0.00%	8 Missing ⚠️
src/util.rs	0.00%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #2241       +/-   ##
===========================================
+ Coverage   30.91%   41.09%   +10.18%     
===========================================
  Files          53       55        +2     
  Lines       20112    20857      +745     
  Branches     9755     9844       +89     
===========================================
+ Hits         6217     8571     +2354     
- Misses       7922     8176      +254     
+ Partials     5973     4110     -1863     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Saruniks]

I'll add montekki as a co-author.

montekki, please let us know if you have any comments on this PR or if you would like to takeover the work.

---

I'm not sure how to prepare this for a merge. Fixing the merge conflicts and getting the CI to pass was straightforward, but there are a lot of changes, and there's a risk for fallout.
I'll give an update after testing this on my setup.

---

I wonder what's up with the freebsd tests though.

[Saruniks]

I did some testing. For some reason sccache_dist scheduler doesn't work properly without SCCACHE_NO_DAEMON=1.

I tested and it even works like this:
SCCACHE_NO_DAEMON=1 sccache-dist scheduler --config /etc/sccache/scheduler.conf &

This is the same reason why freebsd ci doesn't pass. I'll try to fix this issue.

I don't have any useful logs yet.

[Saruniks]

I found out that I need to create a tokio::Runtime instance manually after call to daemonize, instead o using the #[tokio::main] attribute.

[Xuanwo]

All unit test for freebsd passed, we only have a No distributed compilations succeeded. Seems the dist-server or dist-client doesn't start correctly on freebsd.

We expect to see something like:

Statistics of second buildtest
  Compile requests                      13
  Compile requests executed              4
  Cache hits                             4
  Cache hits (Rust)                      4
  Cache misses                           0
  Cache hits rate                   100.00 %
  Cache hits rate (Rust)            100.00 %
  Cache timeouts                         0
  Cache read errors                      0
  Forced recaches                        0
  Cache write errors                     0
  Compilation failures                   0
  Cache errors                           0
  Non-cacheable compilations             0
  Non-cacheable calls                    8
  Non-compilation calls                  1
  Unsupported compiler calls             0
  Average cache write                0.000 s
  Average compiler                   0.000 s
  Average cache read hit             0.000 s
  Failed distributed compilations        0

[Saruniks]

I fixed freebsd test by creating tokio::Runtime manually instead of using #[tokio::main]:

    let runtime = Runtime::new().context("Failed to create Tokio runtime")?;
    runtime.block_on(async { http_scheduler.start().await })?;

It was a problem that I noticed while testing on ubuntu too.

[Xuanwo]

Thank you, @Saruniks, for making this happen, and also thanks to @montekki for the original work. Overall, this PR looks good to me.

Some details inside this PR can be polished in the future.

[Xuanwo]

cc @sylvestre would you like to take another look?

[Saruniks]

Xuanwo I just noticed that native_tls_no_sni_client_builder method might not be needed anymore as seanmonstar/reqwest#1328 was already fixed

[Xuanwo]

Please feel free to remove it, thanks in advance.