Thread (21 messages) flat view 21 messages, 3 authors, 2012-11-21

Re: [PATCH v2] checkpatch: add double empty line check

From: Joe Perches <joe@perches.com>
Date: 2012-11-21 15:01:17
Also in: lkml

On Wed, 2012-11-21 at 11:42 +0200, Eilon Greenstein wrote:
quoted hunk ↗ jump to hunk
On Tue, 2012-11-20 at 15:41 -0800, Joe Perches wrote:
quoted
On Tue, 2012-11-20 at 23:19 +0000, Andy Whitcroft wrote:
quoted
On Tue, Nov 20, 2012 at 01:58:48PM -0800, Joe Perches wrote:
quoted
+# check for multiple blank lines, warn only on the second one in a block
+		if ($rawline =~ /^.\s*$/ &&
+		    $prevrawline =~ /^.\s*$/ &&
+		    $linenr != $last_blank_linenr + 1) {
+			CHK("DOUBLE_EMPTY_LINE",
+			    "One blank line separating blocks is generally sufficient\n" . $herecurr);
+			$last_blank_linenr = $linenr;
+		}
+
 # check for line continuations in quoted strings with odd counts of "
 		if ($rawline =~ /\\$/ && $rawline =~ tr/"/"/ % 2) {
 			WARN("LINE_CONTINUATIONS",
Pretty sure that will fail with combination which have removed lines.
Not as far as I can tell.
Deleted lines followed by inserted lines seem
to work OK.

This check is located after the test that ensures
the current $line/$rawline is an insertion.
But you do not look at the next line, so you will miss something like
that:
diff --git a/test.c b/test.c
index e3c46d4..e1c6ffc 100644
--- a/test.c
+++ b/test.c
@@ -15,7 +15,8 @@
  * something
  * something
  * something
- * next line was already empty */
+ * next line was already empty, but I'm adding another one now*/
+
Hi Eilon.

Thanks for the test case.

That's true, but I'm OK with missing a few cases in the
search for simplicity as long as there aren't significant
false positives.

For instance the next test
# check for line continuations in quoted strings with odd counts of "
		if ($rawline =~ /\\$/ && $rawline =~ tr/"/"/ % 2) {
			WARN("LINE_CONTINUATIONS",

That fails if the rawline is:
	"\"" \
Does it matter much?  Probably not.
I suppose that test could be improved by using $line.

checkpatch isn't a perfect tool.  Given how it's constructed,
I doubt it ever could be.

No doubt you and Andy will find a better solution.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help