Skip to contentSkip to conversation

utf8: use ARRAY_SIZE() in git_wcwidth()

PATCH Superseded Beat Bolli 2 participants 3 messages Oct 11, 2019 – Oct 12, 2019

lore
Share & tools
lore

Beat Bolli

author
#0 lore raw

PATCH 1/1 utf8: use ARRAY_SIZE() in git_wcwidth()

This macro has been available globally since b4f2a6ac92 ("Use #define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))", 2006-03-09), so let's use it.

---
It's idle to speculate about why the macro wasn't used in the original 9e83266525 ("commit-tree: encourage UTF-8 commit messages.", 2006-12-22) or why it hasn't been cleaned up since then. Possible reasons may be the line wrap within the division or the fact that one sizeof() takes the variable while the other one takes the type. Either of these would make a normal search fail.

 utf8.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
utf8.c +2 -4
index 3b42fadffd..164e20c817 100644
@@ -95,13 +95,11 @@ static int git_wcwidth(ucs_char_t ch)
 		return -1;
 
 	/* binary search in table of non-spacing characters */
-	if (bisearch(ch, zero_width, sizeof(zero_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, zero_width, ARRAY_SIZE(zero_width) - 1)
 		return 0;
 
 	/* binary search in table of double width characters */
-	if (bisearch(ch, double_width, sizeof(double_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, double_width, ARRAY_SIZE(double_width) - 1)
 		return 2;
 
 	return 1;
Signature · 1 line
2.21.0.1020.gf2820cf01a
author
#1 lore raw

PATCH v2 1/1 utf8: use ARRAY_SIZE() in git_wcwidth()

This macro has been available globally since b4f2a6ac92 ("Use #define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))", 2006-03-09), so let's use it.

---
Brown paperbag fix. Sorry!

It's idle to speculate about why the macro wasn't used in the original 9e83266525 ("commit-tree: encourage UTF-8 commit messages.", 2006-12-22) or why it hasn't been cleaned up since then. Possible reasons may be the line wrap within the division or the fact that one sizeof() takes the variable while the other one takes the type. Either of these would make a normal search fail.

 utf8.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
utf8.c +2 -4
index 3b42fadffd..5c8f151f75 100644
@@ -95,13 +95,11 @@ static int git_wcwidth(ucs_char_t ch)
 		return -1;
 
 	/* binary search in table of non-spacing characters */
-	if (bisearch(ch, zero_width, sizeof(zero_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, zero_width, ARRAY_SIZE(zero_width) - 1))
 		return 0;
 
 	/* binary search in table of double width characters */
-	if (bisearch(ch, double_width, sizeof(double_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, double_width, ARRAY_SIZE(double_width) - 1))
 		return 2;
 
 	return 1;
Signature · 1 line
2.21.0.1020.gf2820cf01a
#2 lore raw

utf8.c static int git_wcwidth(ucs_char_t ch) · line 98

Quoted 11 lines · Beat Bolli
utf8.c +1 -2
index 3b42fadffd..5c8f151f75 100644
@@ -95,13 +95,11 @@ static int git_wcwidth(ucs_char_t ch)
 		return -1;
 
 	/* binary search in table of non-spacing characters */
-	if (bisearch(ch, zero_width, sizeof(zero_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, zero_width, ARRAY_SIZE(zero_width) - 1))

I wondered if we want a wrapper similar to QSORT() macro so that we do not have to repeat zero_width (and double_width below) like this, but bisearch() itself is local to this file and there are only these two callers, so it probably is not worth it.

The original is waiting for a bug when zero_width changes its type, by not dividing with sizeof(*zero_width). The use of ARRAY_SIZE() is quite appropriate here.

Thanks.

Quoted 9 lines
utf8.c +1 -2
 		return 0;
 
 	/* binary search in table of double width characters */
-	if (bisearch(ch, double_width, sizeof(double_width)
-				/ sizeof(struct interval) - 1))
+	if (bisearch(ch, double_width, ARRAY_SIZE(double_width) - 1))
 		return 2;
 
 	return 1;