[PATCH] D80950: [clang-format] [PR44542,38872] String << String always get a forced newline.

[Sam McCall via Phabricator via cfe-commits]

sammccall added a comment.

IMO this *is* actually a good heuristic: `<<` between string literals indicates some intent to break, and when the stream is entirely literals the formatting is good.
(@krasimir threw this into our regression testing, and where the formatting was different, it was always worse than before).

e.g. before:

  LOG(ERROR) << "(" << foo << ")"
              << " The printer " << bar << " is on fire.";

after:

  LOG(ERROR) << "(" << foo << ")" << " The printer " << bar
             << " is on fire.";

or before:

  LOG_IF(DFATAL, exp_await_calls < num_await_calls_)
      << "Equality condition can never be satisfied:"
      << " exp_await_calls=" << exp_await_calls
      << " num_await_calls_=" << num_await_calls_;

after:

  LOG_IF(DFATAL, exp_await_calls < num_await_calls_)
      << "Equality condition can never be satisfied:" << " exp_await_calls="
      << exp_await_calls << " num_await_calls_=" << num_await_calls_;

However the fact that we can't always infer the intent to break leads to some annoying regularities, and it's hard to imagine heuristics could catch every case. (I'm thinking about looking at leading/trailing characters in string literals, but that's going to be really confusing when it goes wrong).

I guess it's not the end of the world to lose this behavior, but would certainly prefer to keep it at least for google style.

(Incidentally, the fact that the google style guide leans towards using iostreams mostly only for logging-type stuff might be relevant to the types of examples that are typical)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D80950/new/

https://reviews.llvm.org/D80950