editoast: refactor infra tests part 7

[hamz2a]

Part of #6980

It's better to review the PR commit by commit.
I created intermediate functions {function}_v2 to facilitate the migration, but I deleted everything at the end.

I also refactored persist_railjson to use DbConnection directly instead of DbConnectionPool. Consequently, I removed the macro persist! that was in there and futures::try_join!, and I added a transaction.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

Attention: Patch coverage is 82.45614% with 30 lines in your changes missing coverage. Please review.

Project coverage is 28.21%. Comparing base (bea5cee) to head (6b5fd5a).
Report is 7 commits behind head on dev.

Files	Patch %	Lines
editoast/src/views/infra/edition.rs	59.37%	13 Missing ⚠️
editoast/src/views/infra/mod.rs	66.66%	7 Missing ⚠️
editoast/src/views/infra/railjson.rs	54.54%	5 Missing ⚠️
editoast/src/views/infra/routes.rs	57.14%	3 Missing ⚠️
editoast/src/views/infra/auto_fixes/mod.rs	77.77%	2 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff              @@
##                dev    #7932      +/-   ##
============================================
- Coverage     28.25%   28.21%   -0.05%     
  Complexity     2075     2075              
============================================
  Files          1276     1276              
  Lines        156193   156314     +121     
  Branches       3084     3084              
============================================
- Hits          44136    44102      -34     
- Misses       110216   110371     +155     
  Partials       1841     1841              

Flag	Coverage Δ
core	75.03% <ø> (ø)
editoast	70.79% <82.45%> (-0.23%)	⬇️
front	9.91% <ø> (-0.01%)	⬇️
gateway	2.34% <ø> (ø)
railjson_generator	87.49% <ø> (ø)
tests	72.93% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[woshilapin]

I'm curious about why we do need a lot of #[serial_test::serial] in commit 3040c34.

Also, I'd like to be convinced that removing futures::try_join_all!() is fine. But to me, it should affect the time of execution for persist_railjson unless I'm missing something.

Apart from that, great work in separating commits in this PR. Thank you, it really eased the review for this big PR.

[hamz2a]

I'm curious about why we do need a lot of #[serial_test::serial] in commit 3040c34.

I added #[serial_test::serial] because I was getting deadlock errors.

[woshilapin]

I'm curious about why we do need a lot of #[serial_test::serial] in commit 3040c34.

I added #[serial_test::serial] because I was getting deadlock errors.

We dug a little bit to try to understand. It's a deadlock detected by the Postgresql instance itself. It might be due to concurrence on the generated id for infras since we do create very long transactions when working with infras in tests.

We did not find an obvious solution to work around the problem. Let's keep it like this.

[leovalais]

Thanks for this work! PoolV1 is almost gone! 😄

[leovalais]

Thanks for the changes! A few more and 🚀
I'll test your PR on Monday.

There are commits like merge branch dev or rebase in your history. Those will need to be squashed.

[leovalais]

Great job and thank you for the investigation of the deadlocks bug and for rearranging the history!

Once this is merged we'll need to check that all non-TSV1 endpoints are using DbPoolV2 and that we didn't miss any test that was using TestFixture.

After that we can probably work on making the DbConnection opaque to allow us to remove any diesel from views :)