DBG: add support for pretty printing nonzero numbers

[Kobzol]

This PR adds support for pretty printing NonZero numbers from the standard library. So far I have implemented support for GDB, if you think that my approach is OK, let me know and I'll try to add support for LLDB.

Fixes: #5247

[Kobzol]

Rust 1.32.0 doesn't seem to know about NonZeroI32, that's why the test is failing. Should I change the test to only include NonZeroU32?

[Undin]

@Kobzol some time ago we introduced special comments to restrict version of rustc that should be used for a particular test
They are // min-version: %rustc-version% or // max-version: %rustc-version%. You can find them, for example, here
So you may restrict minimal version of rustc for NonZero* test. Or you may split it into several tests and specify min rustc version only for part of them

[Kobzol]

Thank you, I have added a minimum rustc version comment.

When I added tests for LLDB, I have noticed something peculiar: they immediately passed, even though I have not implemented the LLDB formatter for NonZero integers yet. As you can see in my latest commit (c55a3a7), I have purposefully added an error to a different LLDB test and CI failed, as expected, so something is wrong with my setup.
I'm running the tests locally with ./gradlew runPrettyPrintersTests --no-daemon. Do you have any idea what might be causing the problem?

[artemmukhin]

@Kobzol I'm not sure what might be wrong with your setup. You can try to debug pretty_printers_tests by running pretty_printers_tests/main.rs with the following command in the run configuration:

run --package pretty_printers_test --bin pretty_printers_test -- lldb "<IntelliJ_Rust>/deps/clion-2020.1/bin/lldb/linux/lib/python3.6/site-packages" <193 or 201>

[Kobzol]

Thank you. I didn't notice before that LLDB needs special syntax for checking the output in 201 -- this was the problem. I also used a wrong minimum rustc version. The tests should now be fixed (I hope :)
).

[Kobzol]

Damn, the tests failed on my fork's Github actions :-( I use the exact same // min-version: 1.X.Y annotation that is used in other tests that are successfully skipped.

I thought that this annotation will skip compilation of the test, but it seems that it only skips checking the test debug output, is that correct? In that case I have to modify the test to work with Rust 1.32, otherwise the tests will not pass.

[artemmukhin]

@Kobzol Oh, you're right. You can try changing test_runner.rs to compile the test after parsing & checking debugger commands (i.e. to call compile_test after calling parse_debugger_commands).

[Kobzol]

Thanks, now the tests finally pass. They are possibly also just a teeeeny tiny bit faster now that skipped tests are not being compiled :)

[artemmukhin]

LGTM. Let's squash the commits into two (one about NonZero and another one about test_runner.rs) before merge

[artemmukhin]

Thank you!

bors r+

[bors]

Build succeeded:

• continuous-integration/appveyor/branch
• continuous-integration/travis-ci/push