On 1/14/22 10:56 AM, Ján Tomko wrote:
On a Friday in 2022, Tim Wiederhake wrote:
This was not mentioned before.
Signed-off-by: Tim Wiederhake <twiederh@xxxxxxxxxx>
---
docs/coding-style.rst | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/docs/coding-style.rst b/docs/coding-style.rst
index 14c5136398..e1ed34f764 100644
--- a/docs/coding-style.rst
+++ b/docs/coding-style.rst
@@ -600,6 +600,19 @@ calling another function.
...
}
+Define variables on separate lines. This allows for smaller, easier to
+understand diffs when changing them. Define variables in the smallest
+possible scope.
+
+::
+
+ GOOD:
+ int x;
+ int y;
+
+ BAD:
+ int x, y;
+
Please use longer variable names and initialize some too, to illustrate
it better, e.g.:
int count = 0, nnodes;
Personally I don't mind:
size_t i, j;
that much - even though removing one does cause churn, they are simple
to read.
I also don't mind combining simple things like that, but am willing to
go full-isolated just for consistency's sake.
Since it's Friday and we're talking about personal preferences - I
personally dislike the use of i and j (and anything else with a single
letter) as variable names, because it makes using a text search for
occurences pointless. Sure, longer variable names could also be a
substring of something else, and any variable could be re-used
elsewhere, but even then a search is mildly usable.
(On the other hand, sometimes a loop is just a loop and it takes too
much brain capacity to think of a meaningful name for the index. I used
to work with someone who always used "ii" and "jj" for generic loop
indexes because they were then easy to search for with few false
positives (well - "ascii", "skiing", and a surprisingly high number of
other more obscure words, but still...) , and I internalized that
practice myself. After having libvirt patches with that rejected a
couple times, I unlearned and conformed to the hive :-))
