Re: [PATCH] config: add support for --bool and --int while setting values

Frank Lichtenheld [off-list ref] writes:

On Mon, Jun 25, 2007 at 04:06:34PM +0200, Johannes Sixt wrote:

quoted

Frank Lichtenheld wrote:

quoted

Signed-off-by: Frank Lichtenheld <redacted>

Please excuse if I'm missing the big picture, but why do we need this
change?

- Of course the user or script calling git-config can do the
  normalization and error checking, if they want to. But I would
  prefer to have it available in git-config.
- I would prefer that these options wouldn't be silently ignored,
  because that can be confusing (at least it is documented now, but
  still). So we should either using them or error out. I prefer the former.

Something that I forgot to mention in the previous mail:
One real problem with the patch is that it expands the k,m,g suffixes
for integer values. It probably shouldn't do that.

How about doing something like this, then?

git_config_int() knows that a missing value is a nonsense and
barfs on such an input, so (value ? value : "") is redundant
here.  Besides, you check value == NULL much earlier in this
function.

diff --git a/builtin-config.c b/builtin-config.c
index 9973f94..33b60ec 100644
--- a/builtin-config.c
+++ b/builtin-config.c

@@ -148,10 +148,11 @@ char* normalize_value(const char* key, const char* value)
 	if (type == T_RAW)
 		normalized = xstrdup(value);
 	else {
-		normalized = xmalloc(64);
-		if (type == T_INT)
-			sprintf(normalized, "%d",
-				git_config_int(key, value?value:""));
+		normalized = xmalloc(64 + strlen(value));
+		if (type == T_INT) {
+			int v = git_config_int(key, value);
+			sprintf(normalized, "%d # %s", v, value);
+		}
 		else if (type == T_BOOL)
 			sprintf(normalized, "%s",
 				git_config_bool(key, value) ? "true" : "false");