type-number: fix clang-tidy modernize-use-nullptr

[ViniciusCestarii]

Prevent warning --modernize-use-nullptr when using clang-tidy on downstream projects (Noticed when pushing 2140-dev/bitcoin#42, this change was also necessary for bitcoin/bitcoin#29409 so CI doens't error)

[DrahtBot]

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Reviews

See the guideline for information on the review process.

Type	Reviewers
Stale ACK	ryanofsky

If your review is incorrectly listed, please copy-paste <!--meta-tag:bot-skip--> into the comment that the bot should ignore.

[maflcko]

Looks like CI fails in mptest, which may be unrelated.

Also, I wonder why the CI passes fine without this change, when it does claim to run clang-tidy:

libmultiprocess/ci/configs/llvm.bash

Line 8 in 61de697

-DMP_ENABLE_CLANG_TIDY=ON

[ViniciusCestarii]

Also, I wonder why the CI passes fine without this change, when it does claim to run clang-tidy:

The CI does run clang-tidy, but it only warns about code that actually gets compiled. The float overload of CustomReadField (the one with = 0) is a template, so it only turns into real code when something uses a float type. libmultiprocess's own foo.capnp has no Float64 field, so that code is never generated and clang-tidy never looks at it. Bitcoin Core's .capnp does use Float64, so mpgen generates the code, and then clang-tidy can see the type and warn with modernize-use-nullptr.

Adding a Float64 field on foo.capnp and foo.h would have triggered this error here. I believe there other types that are not being tested too.

[ViniciusCestarii]

Looks like CI fails in mptest, which may be unrelated.

The same thing happened on #278 which only changes a markdown file. I will give a look on what's happening

[ryanofsky]

Code review ACK 91a951f. Thanks for the PR. It could also be a good to add a test that would trigger this error. I guess would just need to add a test passing a floating point value.

[maflcko]

Added test in #294