BUG/MINOR: pattern: pat_ref_set: return 0 if err was found

pat_ref_set_elt() returns 0, if we are run out of memory or can't parse a new
map value. Any arror message emitted by pat_ref_set_elt() is saved in err
buffer, if its provided by caller. These error messages are cumulated during
the loop.

pat_ref_set() is used to update values in map, referred to the same given key.
If during the update pat_ref_set_elt() fails, let's retun 0 to caller
immediately. We have the same non-unique key and the same new value in each
loop. So it seems quite odd to cumulate the same error messages and print it in
CLI:

        > add map @1 mytest.map <<
        + 1.0.1.11 TestA
        + 1.0.1.11 TESTA
        + 1.0.1.11 test_a
        +

        > set map mytest.map 1.0.1.11 15
         unable to parse '15' unable to parse '15' unable to parse '15'.

cli_parse_set_map(), which calls pat_ref_set() to update map, will return only
one error message with this patch:

> set map mytest.map 1.0.1.11 15
 unable to parse '15'.

hlua_set_map() and http_action_set_map() don't provide error buffer and will
just exit on the first error.

This should be backported in all stable versions.

(cherry picked from commit 911f4d93d436a9067fb45168f70f79b5916489b0)
Signed-off-by: Christopher Faulet <cfaulet@haproxy.com>

diff --git a/src/pattern.c b/src/pattern.c
index da81e17..90f386e 100644 (file)
--- a/src/pattern.c
+++ b/src/pattern.c
@@ -1792,6 +1792,7 @@ int pat_ref_set(struct pat_ref *ref, const char *key, const char *value, char **
                if (!pat_ref_set_elt(ref, elt, value, &tmp_err)) {
                        memprintf(err, "%s, %s", err && *err ? *err : "", tmp_err);
                        ha_free(&tmp_err);
+                       return 0;
                }
                found = 1;
        }