Turn off DeterministicTrivial test (see #8356)

[ffaf1]

See #8356

---

Please include the following checklist in your PR:

• Patches conform to the coding conventions.
• no relevant info to users
• The documentation has been updated, if necessary.

Please also shortly describe how you tested your change. Bonus points for added tests!

[ulysses4ever]

I assume we won't be waiting for two days here?

[ffaf1]

Ambiguous module name ‘Crypto.Hash.SHA256’:
      it was found in multiple packages:
      cryptohash-sha256-0.11.102.1 cryptohash-sha256-0.11.102.1

??? I keep looking at the two cryptohash, they are the same!

[Mikolaj]

??? I keep looking at the two cryptohash, they are the same!

That's the very test we were trying to disable. Apparently it fails before our disabling kicks in. Perhaps the skipIf needs to be moved to the start of the test.

[ffaf1]

Putting skip... on top still gives me that weird import error.

With this, it works, but it is ugly (had to comment BS16 import) and leaves all the questions on why it fails open.

import Test.Cabal.Prelude
import qualified Data.ByteString as BS
-- import qualified Data.ByteString.Base16 as BS16
import qualified Crypto.Hash.SHA256 as SHA256
import System.FilePath
    ( (</>) )

main = cabalTest $ do

    cabal "v2-sdist" ["deterministic"]
    env <- getTestEnv
    let dir = testCurrentDir env
        knownSdist = dir </> "deterministic-0.tar.gz"
        mySdist = dir </> "dist-newstyle" </> "sdist" </> "deterministic-0.tar.gz"

    -- This helps to understand why this test fails, if it does:
    --
    -- shell "tar" ["-tzvf", knownSdist]
    -- shell "tar" ["-tzvf", mySdist]
    --

    known <- liftIO (BS.readFile knownSdist)
    unknown <- liftIO (BS.readFile mySdist)

    -- Not using `expectBroken` becuase the test succedes locally but
    -- fails in CI.
    skipIf "#8356" True
    -- assertEqual "hashes didn't match for sdist" (BS16.encode $ SHA256.hash known) (BS16.encode $ SHA256.hash unknown)
    assertEqual "hashes didn't match for sdist" True True

[Mikolaj]

Oh, wow, so it fails in the import? No wonder we can't put skipIf far enough up. Let's comment out the import in that case!

I wonder what uncommenting This helps to understand why this test fails, if it does would show, but that's for another time.

[Mikolaj]

And actually, it would probably not show anything useful, because the assertion probably doesn't fail, it's the import that fails.

[ffaf1]

The last iteration:

• works;
• clearly states the test is broken in unfathomable ways;
• marks it as skipped lest we forget about it.

[ffaf1]

Notice how on different platforms, we get different imports error (on a linux virtual machine, BS16; on Windows CI, SHA256).

What a weird behaviour.

[Mikolaj]

Emergency-merging to unblock CI for our devs.

[Mikolaj]

Unfortunately, this needs to be backported, because it blocks CI on branch 3.8 as well.

[Mikolaj]

@mergify backport 3.8

[mergify]

backport 3.8

✅ Backports have been created

• #8369 Turn off DeterministicTrivial test (see #8356) (backport #8357) has been created for branch 3.8