Clang-format configuration

[offa]

Description

Initial .clang-format file, open for discussion. The config doesn't contain much as the defaults fit most cases already. Documentation of all options is available here.

Settings to discuss / change

• ❓ NamespaceIndentation: Indention of namespaces – All, Inner or None?
• ❓ IndentPPDirectives: Indention of proprocessor directives

GitHub Issues

#1182

[codecov]

Codecov Report

Merging #1920 into master will increase coverage by 1.80%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1920      +/-   ##
==========================================
+ Coverage   86.80%   88.59%   +1.80%     
==========================================
  Files         131      138       +7     
  Lines        3923     5619    +1696     
==========================================
+ Hits         3405     4978    +1573     
- Misses        518      641     +123     

[horenmar]

Right, I'll try to find some time to test the configuration out.

However, since opening #1182, I started having some misgivings about clang-format. From my experience with other projects, it doesn't handle some constructs well, and the actual formatting depends on the specific version of clang-format, making inconsistencies likely.

[offa]

Please let me know if there are some issues. The result might not be perfect, but maybe we can get some consistency.

[horenmar]

So, I spent some time playing with the config and made some further changes (pushed them as a separate commit). There is still at least one problem that needs fixing before I am willing to say "use clang-format now". See below

Some notes on the current settings

• NamespaceIndentation - I would love a hybrid between Inner and All, where the most-nested namespace is indented, but the less nested ones aren't. That doesn't exist, so I ended up going with All.
• IndentPPDirectives - This is currently inconsistent in the codebase, but AfterHash is the style I prefer for new code.
• AllowShortEnumsOnASingleLine - preferably would be set to False, but that seems to be a clang-format-11 thing.

The big problem is with SpacesInParentheses. I don't care for it that much personally, but the absolute majority of current code is written in style that is closely resembled by using it, but not quite. The problems come from parenthesized expressions (such as defensively written macro argument expansion), which aren't either function decl or function call.

I think I will end up turning it on and trying it out a bit more, because I don't want to generate mostly pointless diff against all code.

[offa]

AllowShortEnumsOnASingleLine - preferably would be set to False, but that seems to be a clang-format-11 thing.

I've just check the clang docs, you are right, clang-format 10 doesn't mention it.

The problems come from parenthesized expressions (such as defensively written macro argument expansion), which aren't either function decl or function call.

I see, in the worst case it might be an option to disable clang format for particular sections.

I think I will end up turning it on and trying it out a bit more, because I don't want to generate mostly pointless diff against all code.

👍