Wietse Venema:
> > * Wietse Venema <postfix-users@postfix.org>:
> > > Victor Duchovni:
> > > > On Tue, Jan 18, 2011 at 03:56:45PM -0500, Wietse Venema wrote:
> > > >
> > > > > Something along the lines of:
> > > > >
> > > > > /*
> > > > > * Workaround. The "," was already in use as dnsbl list separator.
> > > > > */
> > > > > for (keep = 0, cp = var_psc_dnsbl_sites; *cp; cp++) {
> > > > > if (*cp == '[') {
> > > > > keep++;
> > > > > } else if (*cp == ']' && keep > 0) {
> > > > > keep--;
> > > > > } else if (*cp == ',' && keep <= 0) {
> > > > > *cp = ' ';
> > > > > }
> > > > > }
> > > > > dnsbl_site = argv_split(var_psc_dnsbl_sites, ", \t\r\n");
> > > >
> > > > Right, reasonably elegant, but with the split now on just " \t\r\n".
> > >
> > > But having "," inside an access control feature it is likely to
> > > break third-party tools that maintain Postfix configuration files.
> > >
> > > The alternative is to change the address filter syntax, and to
> > > replace "," by a different set separator such as ";".
>
> The [x,x] syntax never worked in main.cf, so I can't break
> configuration file compatibility by changing to [x;x].
>
> But I would have to update a dozen regression tests that I wrote
> for the code module that implements the address filter.
This changes the syntax to:
potscreen_dnsbl_sites:
Specify a list of domain=filter*weight entries, separated by comma or
whitespace.
o When no "=filter" is specified, postscreen(8) will use any non-
error DNSBL reply. Otherwise, postscreen(8) uses only DNSBL
replies that match the filter. The filter has the form d.d.d.d,
where each d is a number, or a pattern inside [] that contains
one or more ";"-separated numbers or number..number ranges.
...
reject_rbl_client rbl_domain=d.d.d.d
Reject the request when the reversed client network address is
listed with the A record "d.d.d.d" under rbl_domain (Postfix
version 2.1 and later only). Each "d" is a number, or a pattern
inside "[]" that contains one or more ";"-separated numbers or
number..number ranges (Postfix version 2.8 and later).
...
And likewise for all reject_*bl_* features.
Wietse
