> > However is practice this does not happen and so the mess continues.
>
> I agree with your last clause!
That's why I asked in the beginning. If everbody obeyed the RFCs,
we would have much less of a mess.
I asked for multiple authenticators of one mechanism that would be tried
until one suceeds. This turned out easier than I thought, the patch
against 3.20 is appended. Without this patch, you have to squeeze all
lookups into one authenticator. I screwed up indentation, but that
makes the diff easier to read.
Philip: Any chance to get this or an equivalent patch into the next release? :)
It really helps me a lot.
Michael
----------------------------------------------------------------------
--- src/smtp_in.c.orig Wed Jan 10 15:38:50 2001
+++ src/smtp_in.c Wed Jan 10 15:44:48 2001
@@ -1816,6 +1816,7 @@
#ifdef HAVE_AUTH
auth_instance *au;
int c;
+ int foundAuthenticator;
#endif
switch(smtp_read_command())
@@ -1872,17 +1873,12 @@
/* Search for an authentication mechanism which is configured for use
as a server. */
- for (au = auths; au != NULL; au = au->next)
+ for (foundAuthenticator = 0, au = auths; au != NULL; au = au->next)
{
- if (strcmpic(s, au->public_name) == 0 && au->server) break;
- }
-
- if (au == NULL)
- {
- smtp_printf("504 %s mechanism not supported\r\n", s);
- break;
- }
+ if (strcmpic(s, au->public_name) == 0 && au->server)
+ {
+ foundAuthenticator=1;
/* Run the checking code, passing the remainder of the command
line as data. Initialize $0 empty. The authenticator may set up
other numeric variables. If authentication succeeds, expand the
@@ -1913,6 +1909,16 @@
}
expand_nmax = -1; /* Reset numeric variables */
+
+ if (c!=FAIL) break;
+ }
+ }
+ if (!foundAuthenticator)
+ {
+ smtp_printf("504 %s mechanism not supported\r\n", s);
+ break;
+ }
+
switch(c)
{
