Re: tflags publish <= What is this good for? (Re: Rule for: Subject contains email address from To: header?)

On 2026-06-01 at 09:08:33 UTC-0400 (Mon, 01 Jun 2026 09:08:33 -0400)
Bill Cole <[email protected]>
is rumored to have said:

> On 2026-06-01 at 07:56:41 UTC-0400 (Mon, 1 Jun 2026 13:56:41 +0200)
> Benoit Panizzon <[email protected]>
> is rumored to have said:
>
>> Hi
>>
>> I figured out the culpit why a rule did not work was tflags publish
>
> That's obviously not good.
>
> The 'publish' tflag exists to force a rule in the full set of rules in =
our collection into the active list that is distributed.
>
>> Actual finding in trying to get the shipped rule TO_IN_SUBJ to match
>> anything. Just created a new rule with one of the regexp:
>>
>> header          TEST_EMAIL_SUBJ_TO      ALL =3D~ /\nSubject:(?=3D[^\n]=
{0,200}@)[^\n]{0,200}([a-z][a-z0-9_.]{3,80}@(?:[a-z0-9_]{1,80}\.){1,4}[a-=
z]{2,30})(?:[^\n]+\n)*To:\s+[^\n]{0,100}\1[^a-z0-9.]/ism
>> describe        TEST_EMAIL_SUBJ_TO      Testing: Email in Subject then=
 To
>> score           TEST_EMAIL_SUBJ_TO      1
>> tflags          TEST_EMAIL_SUBJ_TO      publish
>>
>> When I remove the tflags line, the rule is run and matches as
>> expected when invoking SpamAssassin from MIMEDefang.
>>
>> When I keep the tflags line, then the rule seems to only run when
>> calling SpamAssassin from command line.
>
> This does not make sense. The publish tflag should not have any effect =
on how messages are checked.
>
> As described, this is definitely a bug.

However, I cannot reproduce it. I constructed a message matching TO_IN_SU=
BJ and ran it through both MIMEDefang and the spamassassin script. It hit=
 in both cases.

Please note that TO_IN_SUBJ is not a simple rule. It is a meta-rule and w=
ill not fire if the message appears to be from a discussion list OR if th=
e envelope sender matches the From header address.

>> So I wonder if there might be some setting somewhere to cause
>> spamassassin not running rules with a specific tflags in some
>> situations?

Thankfully, no.

The next step is for you to validate your MIMEDefang configuration to mak=
e sure that it is loading the same rules as the spamassassin script. Note=
 that by default, MD loads /etc/mail/sa-mimedefang.cf, which the spamassa=
ssin script does not load. w


-- =

 Bill Cole
 [email protected] or [email protected]
 (AKA @[email protected] and many *@billmail.scconsult.com addresses=
)
 Please keep discussion mailing list replies *on-list*
 Not Currently Available For Hire