[patch] ldb_msg_add_string()

[tridge at samba.org]

Simo,

 > well actually I am more concerned about things like:
 > 
 >    ldb_msg_add_empty(ldb, "comment", LDB_FLAG_MOD_REPLACE);
 > 
 > and then forgetting to really set the new value.

that's true, but it's not relevant for my patch :-)

If you do the above, and forget to set the value, then the check in
ldb_msg_sanity_check() will still kick in. I'm not proposing to change
that check, I'm just proposing to apply exactly the patch I posted
(the change to ldb_msg_add_string()).

 > See the ldb_msg_add_empty() example.

lets see if you can find another example :-)

 > I have to think about this, maybe we can silently discard an empty
 > attribute but give out a level 0 debug error.
 > That might be a good enough compromise.

hmmm, no, I don't think thats a good idea.

Maybe you could look at my patch again? Perhaps you didn't notice that
it doesn't actually add an empty attribute to the message?

Cheers, Tridge

PS: Here it is again for easy reference:

Index: common/ldb_msg.c
===================================================================
--- common/ldb_msg.c    (revision 18835)
+++ common/ldb_msg.c    (working copy)
@@ -227,6 +227,11 @@
        val.data = discard_const_p(uint8_t, str);
        val.length = strlen(str);

+       if (val.length == 0) {
+               /* allow empty strings as non-existant attributes */
+               return LDB_SUCCESS;
+       }
+
        return ldb_msg_add_value(msg, attr_name, &val);
 }