Some points on clang-format usage

[Stephan Bergmann]

On 11/17/2017 12:19 PM, Miklos Vajna wrote:
> On Fri, Nov 17, 2017 at 11:16:13AM +0100, Stephan Bergmann <sbergman at redhat.com> wrote:
>> ...which might be reason enough to still revisit the .clang-format rule that
>> causes leading commas, if that wrecks such havoc with comments?
> 
> I think BreakConstructorInitializersBeforeComma is the relevant key.
> Setting it to false would not be controversial, it would just mean we
> leave that area alone, I believe.

So among 
BreakConstructorInitializers:{BeforeColon,BeforeComma,AfterColon} and 
ConstructorInitializerAllOnOneLineOrOnePerLine:{false,true} there 
appears to be no setting that both (a) consistently places each 
initializer on a line of its own (the current behaviour IIUC; beneficial 
for small diffs when modifying an initializer list) and (b) puts the 
comma after the initializer (which would be beneficial for preserving 
trailing comment placement).

Shrug.