ESC / Rome discussion ...

[Stephan Bergmann]

On 10/27/2017 04:49 PM, Jan Holesovsky wrote:
> But still, I see the Thorsten's point why it would be easier for people
> in many cases; that's why I proposed the 'automatic, but ending up as
> an additional changeset' way, that at least gives a chance to inspect &
> do something about that.

I'd propose to have three things:

1  A pre-commit--hook version of the clang-format check (that prevents 
the commit if it fails) that runs iff clang-format is found locally. 
(With the assumption that "core" committers will have clang-format 
installed locally, especially if they commit directly without going via 
Gerrit.)

2  A Gerrit version of the clang-format check that is run for every 
Gerrit patch set, and sets some "Code-Style: -1/+1" Gerrit flag as 
discussed in David's recent mail ("Gerrit Code-Style verification -- 
was ESC / Rome discussion ...").  (For the cases that pass through 
Gerrit and that (1) above misses, like changes 
from---casual---committers that do not have clang-format installed 
locally, or direct edits in the Gerrit web UI.)

3  A Gerrit web UI button that allows to run clang-format on a specific 
Gerrit change and produces a new patch set if necessary (somewhat 
similar to the existing rebase button; useful for---casual---committers 
that do not have clang-format installed locally, if (2) above marked 
their change as "Code-Style: -1".)