While we all understand that excessive use of ternary operator may worsen code readability (e.g. nested, multi-line expression), there are few cases where using it actually improves code readability. For instance, when a function takes a long list of arguments out of which one depends on a boolean expression, or when formatting "yes"/"no" or "on"/"off" values based on a boolean variable (although one can argue that the latter is a subset of the former). Just consider alternatives to: virBufferAsprintf(buf, "<elem>%s</elem>\n", boolVar ? "yes" : "no"); In fact, this pattern occurs plenty in our code. Exempt it from our "no ternary operators" rule. Signed-off-by: Michal Privoznik <mprivozn@xxxxxxxxxx> Reviewed-by: Claudio Fontana <cfontana@xxxxxxx> --- v2 of: https://listman.redhat.com/archives/libvir-list/2022-July/233150.html diff to v1: - Changed wording, as suggested by Daniel. docs/coding-style.rst | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/docs/coding-style.rst b/docs/coding-style.rst index bf0a80fbc5..81bd4474f1 100644 --- a/docs/coding-style.rst +++ b/docs/coding-style.rst @@ -470,7 +470,9 @@ Pointer comparisons may be shortened. All long forms are okay. if (!foo) # or: if (foo == NULL) New code should avoid the ternary operator as much as possible. -Specifically it must never span more than one line or nest: +Its usage in basic cases is warranted (e.g. when deciding between +two constant strings), however, it must never span more than one +line or nest. :: @@ -481,6 +483,9 @@ Specifically it must never span more than one line or nest: char *foo = bar ? bar->baz ? bar->baz->foo : "nobaz" : "nobar"; + GOOD: + virBufferAsprintf(buf, "<element>%s</element>\n", boolVar ? "yes" : "no"); + Preprocessor ------------ -- 2.35.1
