AudioTaperPotTest.ScaleTest fails, but values are equal

I think we need to replace ASSERT_EQ by ASSERT_DOUBLE_EQ in line 23

A PR is here:
https://github.com/mixxxdj/mixxx/pull/1979

Can you confirm that this fixes your issue?

@daschuer, thanks for the quick reply.
Unfortunately, even after applying the patch the same error was triggered for me :-/

Ok, so we need a wider ellipsis.
I will change the code to something like
abbs(a-b) < 0.00001
It it would be interesting how big the difference is though.

I have update the PR for
ASSERT_NEAR((db2ratio(maxDB), catpb.parameterToValue(1), 1e-10);

Please check.

Thank you.

Will do!
You have introduced a leading '(' in that commit btw.

After removing the superfluous '(' from the commit, I still get this:

```
[----------] 1 test from AudioTaperPotTest
[ RUN ] AudioTaperPotTest.ScaleTest
Loading resources from "/build/mixxx/src/mixxx-2.2.0/res/"
src/test/audiotaperpot_test.cpp:44: Failure
Value of: catpb.parameterToValue(1)
  Actual: 1.99526
Expected: db2ratio(maxDB)
Which is: 1.99526
[ FAILED ] AudioTaperPotTest.ScaleTest (1 ms)
[----------] 1 test from AudioTaperPotTest (1 ms total)
```

please change it to:
ASSERT_EQ(db2ratio(maxDB) - catpb.parameterToValue(1), 0);

and post the output.

It looks like the compiler optimizes some fractional places away. Maybe we suffer a float precision overflow. Which compiler version do you use?
Do we have a bug? Normally ASSERT_DOUBLE_EQ should work. mmm

Alright, here it is:

```
[----------] 1 test from AudioTaperPotTest
[ RUN ] AudioTaperPotTest.ScaleTest
Loading resources from "/build/mixxx/src/mixxx-2.2.0/res/"
src/test/audiotaperpot_test.cpp:23: Failure
Value of: 0
Expected: db2ratio(maxDB) - catpb.parameterToValue(1)
Which is: -2.22045e-16
[ FAILED ] AudioTaperPotTest.ScaleTest (0 ms)
[----------] 1 test from AudioTaperPotTest (0 ms total)
```

On Arch Linux we're currently at gcc 8.2.1 (should've mentioned that earlier, sorry).

Thank you.
It is strange because abs(-2.22045e-16) is smaller than 1e-10.

IMHO this should work:
ASSERT_NEAR((db2ratio(maxDB), catpb.parameterToValue(1), 1e-10);

Do you have an idea what went wrong?

I can't reproduce the original failing test on Ubuntu 18.10.
I built with the same settings and gcc 8.2.0

I cannot confirm this either so lets wait until one can confirm the issue.

Thank you for your tests.

@dvzrv: did you have more test results for us? See #11.

Any updates?

@daschuer: Sorry, but it took me some time to get back to this. Mixxx is a little extreme to build, as I can't separate building from testing the application. Scons rebuilds everything in every target :(

Anyways, same problem with whatever was proposed in #11:

```
[----------] 1 test from AudioTaperPotTest
[ RUN ] AudioTaperPotTest.ScaleTest
src/test/audiotaperpot_test.cpp:44: Failure
Value of: catpb.parameterToValue(1)
  Actual: 1.99526
Expected: db2ratio(maxDB)
Which is: 1.99526
[ FAILED ] AudioTaperPotTest.ScaleTest (1 ms)
[----------] 1 test from AudioTaperPotTest (1 ms total)
```

Also, our reproducible build pipeline has the same issue as initially reported:
https://tests.reproducible-builds.org/archlinux/community/mixxx/build1.log

@dvzrv: Ah, got it. The second fail was in line 44 not 23.
Now I have fixed some more places by using ASSERT_DOUBLE_EQ,

This allows +-4 integer minimal steps. One minimal steps matches the reported -2.22045e-16,
Please retest.

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/9562