Re: Priorities for meta rules
On Wed, May 25, 2022 at 07:04:53PM +0300, Henrik K wrote: > > Maybe TxRep depends on some SPF stuff, but that's a can of worms I'm not > gonna open. There should not be any reason for someone to add TXREP into a > meta rule, so I think it's ok. And both AWL and TXREP depend on DKIM, but should be ok as mentioned..
Re: Priorities for meta rules
On Wed, May 25, 2022 at 04:57:59PM +0200, giova...@paclan.it wrote: > > I spotted the issue checking DMARC failures but other rules may be affected > as well. I think DMARC was the last plugin that depends on results from other Plugins, should be fixed now. Maybe TxRep depends on some SPF stuff, but that's a can of worms I'm not gonna open. There should not be any reason for someone to add TXREP into a meta rule, so I think it's ok.
Re: Priorities for meta rules
On 5/25/22 16:12, Henrik K wrote: > On Wed, May 25, 2022 at 04:04:33PM +0200, Giovanni Bechis wrote: >> Hi, >> meta rules starts at the lower priority of the rules they are composed by >> (diff for regression test attached). >> >> If you are using a meta rule that mixes network and non network tests the >> meta rule's priority will be -100. >> After some tests, it seems that if you mix DMARC rules (priority 500) with >> other rules, DMARC checks may start earlier then expected. >> Is this just not well documented (at least in Mail::SpamAssassin::Conf) ? > > Meta-rules have not had "priority" since dynamic meta evaluation was > committed in Bug 7735. > > They do in the sense that you _can_ set a priority value on them, but it > only affects any depending rule to be set to that priority. Which should be > useful when shortcircuiting using meta rules. > > If this is the case of DMARC not waiting for SPF and DKIM module, maybe it > can be done with action_depends_on_tags or something. > I spotted the issue checking DMARC failures but other rules may be affected as well. Giovanni OpenPGP_signature Description: OpenPGP digital signature
Re: Priorities for meta rules
On Wed, May 25, 2022 at 05:12:30PM +0300, Henrik K wrote: > On Wed, May 25, 2022 at 04:04:33PM +0200, Giovanni Bechis wrote: > > Hi, > > meta rules starts at the lower priority of the rules they are composed by > > (diff for regression test attached). > > > > If you are using a meta rule that mixes network and non network tests the > > meta rule's priority will be -100. > > After some tests, it seems that if you mix DMARC rules (priority 500) with > > other rules, DMARC checks may start earlier then expected. > > Is this just not well documented (at least in Mail::SpamAssassin::Conf) ? > > Meta-rules have not had "priority" since dynamic meta evaluation was > committed in Bug 7735. > > They do in the sense that you _can_ set a priority value on them, but it > only affects any depending rule to be set to that priority. Which should be > useful when shortcircuiting using meta rules. > > If this is the case of DMARC not waiting for SPF and DKIM module, maybe it > can be done with action_depends_on_tags or something. I will implement similar async method to DMARC as FromNameSpoof has (_check_eval).
Re: Priorities for meta rules
On Wed, May 25, 2022 at 04:04:33PM +0200, Giovanni Bechis wrote: > Hi, > meta rules starts at the lower priority of the rules they are composed by > (diff for regression test attached). > > If you are using a meta rule that mixes network and non network tests the > meta rule's priority will be -100. > After some tests, it seems that if you mix DMARC rules (priority 500) with > other rules, DMARC checks may start earlier then expected. > Is this just not well documented (at least in Mail::SpamAssassin::Conf) ? Meta-rules have not had "priority" since dynamic meta evaluation was committed in Bug 7735. They do in the sense that you _can_ set a priority value on them, but it only affects any depending rule to be set to that priority. Which should be useful when shortcircuiting using meta rules. If this is the case of DMARC not waiting for SPF and DKIM module, maybe it can be done with action_depends_on_tags or something.
