[cfe-dev] Fwd: clang-format line breaks in nested struct initializer
BOUGET Quentin via cfe-dev
cfe-dev at lists.llvm.org
Fri Jan 25 02:17:55 PST 2019
Hello,
I tried out clang-format (version 5.0.1) today and was wondering why with this
configuration:
---
BreakBeforeBinaryOperators: NonAssignment
BreakBeforeBraces: Linux
IndentWidth: 4
UseTab: Never
...
The following formatting happens:
Input | Output
--------------------------------|------------------------------------
|
#include <stdlib.h> | /****************************
| * *
struct point { | * *
int x; | * *
int y; | * *
}; | * Same format *
| * *
struct segment { | * *
struct point start; | * *
struct point end; | * *
}; | ****************************/
|
int | int
main() | main()
{ | {
struct segment segment = { | struct segment segment = {
.start = { | .start =
.x = 0, | {
.y = 0, | .x = 0,
}, | .y = 0,
.end = { | },
.x = 0, | .end =
.y = 2, | {
}, | .x = 0,
}; | .y = 2,
| },
... | },
return EXIT_SUCCESS; | };
} |
| ...
|
| return EXIT_SUCCESS;
| }
|
(notice how braces are placed in the segment's initializer)
The part I cannot figure out is why clang-format adds a line break after
".start =" but not after "struct segment segment =".
(I mean, ideally, I would like clang-format to not add any line breaks at all,
but if it must add some, I would prefer it adds them consistently. =])
---
I do not know if this helps, but I found a post on stackoverflow that suggested
setting "BreakBeforeBinaryOperators" to "All" to fix this.
I was a bit sceptical considering what I thought the parameter controlled but
tried it anyway... turns out it works! (no line breaks added anywhere)
If someone has time, I would also like to know how allowing clang-format to
break lines before the "=" sign (which is what I understand setting
"BreakBeforeBinaryOperators" to "All" does) helps producing the result I was
looking for. If anything, I would have thought it made it more likely that
a line break would be added after "struct segment segment =".
Regards,
Quentin
PS: I already sent this email once a week ago, before I subscribed to the list.
It has been under review by an admin of the list since then. I don't really
mind, but maybe someone should check on this person.
More information about the cfe-dev
mailing list