Fix compatibility with PG16

[df7cb]

• Replace Abs with fabs
• Remove DatumGetPointer where PG16 got stricter typechecking
• Annotate expected/moc100 output files with PG major and architecture bits they are meant for
• Add expected output files for moc100 on PG16

[esabol]

Would it be premature to add PostgreSQL 16 to the Travis CI? Does apt.postgresql.org.sh support PostgreSQL 16 yet? If 16 does work, then I suggest adding it to .travis.yml in this PR.

Just curious, where did Abs() come from in the first place? Is that an older PostgreSQL thing?

[vitcpp]

I'm not sure how the reference output file moc100_x.out is chosen for comparison. Is it performed automatically by test scripts?

[df7cb]

I'm not sure how the reference output file moc100_x.out is chosen for comparison. Is it performed automatically by test scripts?

https://www.postgresql.org/docs/current/regress-variant.html

Would it be premature to add PostgreSQL 16 to the Travis CI? Does apt.postgresql.org.sh support PostgreSQL 16 yet? If 16 does work, then I suggest adding it to .travis.yml in this PR.

Of course PG16 is already supported. I'll add that shortly.

Just curious, where did Abs() come from in the first place? Is that an older PostgreSQL thing?

Abs was removed in PG16.

[vitcpp]

@df7cb Thank you for the explanation. I'm ok with the patch but we have to fix tests for pg16. I'm not sure how to fix it yet. May be to change snapshot to the latest one?

[df7cb]

The problems that the testsuite has are independent from this pull request.

• On Ubuntu bionic 18.04 (now), there is no PG16 since bionic's zstd is too old so I don't provide that on apt.postgresql.org
• On Ubuntu jammy 22.04, all jobs fail because of compiler warnings (-Werror) in other areas of the code
• On Ubuntu focal 20.04 (which I now put into this PR), only PG10 and PG11 fail, again for something that this PR doesn't touch

The issues should be fixed, but in a separate PR. This PR does fix the actual PG16 problems and I already applied it in the packages on apt.postgresql.org.

[vitcpp]

I propose to enable and fix PG16 tests in another PR and then apply this patch. I can do it but I'm on vacation this week with limited access to the internet.

[esabol]

@vitcpp wrote:

I propose to enable and fix PG16 tests in another PR and then apply this patch. I can do it but I'm on vacation this week with limited access to the internet.

This sounds to me like a very good plan.

Thank you, @df7cb, for detailing the issues with the test suite and such.

[dura0ok]

@df7cb I cherry-picked all your commits to my branch, made a small change and the tests passed.
You can see in #41
@vitcpp @esabol what do you think?

[vitcpp]

@df7cb We fixed some warning messages that resulted in test failures. Could you please be so kind to re-sync your branch? Thank you in advance!

[df7cb]

Branch rebased.

[esabol]

Looks good to me! Let's merge. Thank you , @df7cb !