You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@spamassassin.apache.org by Henrik Krohns <he...@hege.li> on 2019/07/03 10:30:26 UTC
Re: svn commit: r1862459 -
/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
There's also some strange dependency for autolearning in Conf.pm functions
maybe_header_only, maybe_body_only (who came up with the function names? so
vague). It seems net flagged metas are ignored for auto-learn scoring.
Dunno if this is good or bad, more stuff to investigate..
Cc'ing dev
On Wed, Jul 03, 2019 at 01:26:11PM +0300, Henrik Krohns wrote:
>
> That should be considered a bug then.
>
> Probably metas should inherit all the flags from rules they use, it's crazy
> to start flagging all meta rules manually.
>
> On Wed, Jul 03, 2019 at 11:21:02AM +0100, Paul Stead wrote:
> > On the weekly mass check when --reuse is used subrules are not submitted. This
> > means if no reuse on the full rule it'll be ignored during a --reuse run
> > because the subrule isn't submitted.
> >
> > On Wed, 3 Jul 2019 at 11:10, Henrik Krohns <[1...@hege.li> wrote:
> >
> >
> > also no need to reuse PDS_HP_HELO_NORDNS, metas do not launch network
> > rules.
> > Reuse needs to be on the actual query launching rule.
> >
> > On Wed, Jul 03, 2019 at 01:09:19PM +0300, Henrik K wrote:
> > >
> > >
> > > __HELO_RECORD needs to be reuse, not PDS_NO_HELO_DNS
> > >
> > >
> > > On Wed, Jul 03, 2019 at 10:04:41AM -0000, [2]pds@apache.org wrote:
> > > > Author: pds
> > > > Date: Wed Jul 3 10:04:41 2019
> > > > New Revision: 1862459
> > > >
> > > > URL: [3]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > Log:
> > > > These depend on net
> > > >
> > > > Modified:
> > > > spamassassin/trunk/rulesrc/sandbox/pds/[4]20_helo.cf
> > > >
> > > > Modified: spamassassin/trunk/rulesrc/sandbox/pds/[5]20_helo.cf
> > > > URL: [6]http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> > /pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > =======================================================================
> > =======
> > > > --- spamassassin/trunk/rulesrc/sandbox/pds/[7]20_helo.cf (original)
> > > > +++ spamassassin/trunk/rulesrc/sandbox/pds/[8]20_helo.cf Wed Jul 3
> > 10:04:41 2019
> > > > @@ -3,20 +3,24 @@ header __HELO_HIGHPROFILE X-Spam-Rela
> > > > meta PDS_HELO_SPF_FAIL SPF_HELO_FAIL && __HELO_HIGHPROFILE
> > > > describe PDS_HELO_SPF_FAIL High profile HELO that fails SPF
> > > > score PDS_HELO_SPF_FAIL 2.0
> > > > +tflags PDS_HELO_SPF_FAIL net
> > > > reuse PDS_HELO_SPF_FAIL
> > > >
> > > > meta PDS_HP_HELO_NORDNS RDNS_NONE && __HELO_HIGHPROFILE
> > > > describe PDS_HP_HELO_NORDNS High profile HELO with no sender rDNS
> > > > score PDS_HP_HELO_NORDNS 1.0
> > > > +tflags PDS_HP_HELO_NORDNS net
> > > > reuse PDS_HP_HELO_NORDNS
> > > >
> > > > ifplugin Mail::SpamAssassin::Plugin::AskDNS
> > > >
> > > > askdns __HELO_RECORD _LASTEXTERNALHELO_ A /./
> > > > +tflags __HELO_RECORD net
> > > >
> > > > meta PDS_NO_HELO_DNS (__HELO_HIGHPROFILE && !__HELO_RECORD)
> > > > describe PDS_NO_HELO_DNS High profile HELO but no A record
> > > > score PDS_NO_HELO_DNS 2.0
> > > > +tflags PDS_NO_HELO_DNS net
> > > > reuse PDS_NO_HELO_DNS
> > > >
> > > > endif
> > > >
> >
> >
> > References:
> >
> > [1] mailto:hege@hege.li
> > [2] mailto:pds@apache.org
> > [3] http://svn.apache.org/viewvc?rev=1862459&view=rev
> > [4] http://20_helo.cf/
> > [5] http://20_helo.cf/
> > [6] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > [7] http://20_helo.cf/
> > [8] http://20_helo.cf/
Re: svn commit: r1862459 - /spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
Posted by Paul Stead <pa...@gmail.com>.
On Wed, 3 Jul 2019 at 11:58, Paul Stead <pa...@gmail.com> wrote:
> Mine and other's initial scoring is done within an instance of Amavis on
> regular mail servers then masschecked with trunk - any adjustments would
> have to be with the ruleqa/masscheck code on sa-vm1
>
> Paul
>
** this is pretty standard and not limited to Amavis. Most people get the
email and initial X-Spam header from a mailserver instance of SA rather
than trunk - this is the case with --net --reuse
Paul
Re: svn commit: r1862459 -
/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
Posted by Henrik K <he...@hege.li>.
If metas inherit net rules, it's the trunk masscheck that matters. The meta
rule hit in Amavis or whatever, masscheck sees it and should automatically
then consider all subrules.
PS. I've also described previously a setup where pristine mail is archive
on arrival and separately processed by dedicated spamd with only reuse rules
run. This way you get a super clean corpus.
On Wed, Jul 03, 2019 at 11:58:25AM +0100, Paul Stead wrote:
> Mine and other's initial scoring is done within an instance of Amavis on
> regular mail servers then masschecked with trunk - any adjustments would have
> to be with the ruleqa/masscheck code on sa-vm1
>
> Paul
>
> On Wed, 3 Jul 2019 at 11:53, Henrik K <[1...@hege.li> wrote:
>
>
> Since reuse is pretty much only used for masscheckers who use trunk, we
> should be able to figure out something..
>
> net tflag is really only used in Check.pm for skipping rules in non-net
> scoreset and waiting for net rule dependencies to finish. Perhaps it would
> be safe to inherit net flag from meta subrules automatically (basically
> this
> is what is already done manually).
>
>
> On Wed, Jul 03, 2019 at 11:32:35AM +0100, Paul Stead wrote:
> > +1 to the crazy bit, it's something I've realised over the weeks of
> looking
> > into the mass check.
> >
> > With reuse flag on masscheck the X-Spam headers are used to generate the
> corpus
> > submission - subrules aren't output in this header so they're not
> submitted.
> >
> > I'm not certain how to fix without requiring full rule list in headers -
> this
> > causes problems across multiple SA instances and versions - Amavis etc
> >
> > On Wed, 3 Jul 2019 at 11:30, Henrik Krohns <[1...@hege.li> wrote:
> >
> >
> > There's also some strange dependency for autolearning in Conf.pm
> functions
> > maybe_header_only, maybe_body_only (who came up with the function
> names?
> > so
> > vague). It seems net flagged metas are ignored for auto-learn
> scoring.
> > Dunno if this is good or bad, more stuff to investigate..
> >
> > Cc'ing dev
> >
> > On Wed, Jul 03, 2019 at 01:26:11PM +0300, Henrik Krohns wrote:
> > >
> > > That should be considered a bug then.
> > >
> > > Probably metas should inherit all the flags from rules they use,
> it's
> > crazy
> > > to start flagging all meta rules manually.
> > >
> > > On Wed, Jul 03, 2019 at 11:21:02AM +0100, Paul Stead wrote:
> > > > On the weekly mass check when --reuse is used subrules are not
> > submitted. This
> > > > means if no reuse on the full rule it'll be ignored during a
> --reuse
> > run
> > > > because the subrule isn't submitted.
> > > >
> > > > On Wed, 3 Jul 2019 at 11:10, Henrik Krohns <[1][2][3]hege@hege.li
> > wrote:
> > > >
> > > >
> > > > also no need to reuse PDS_HP_HELO_NORDNS, metas do not launch
> > network
> > > > rules.
> > > > Reuse needs to be on the actual query launching rule.
> > > >
> > > > On Wed, Jul 03, 2019 at 01:09:19PM +0300, Henrik K wrote:
> > > > >
> > > > >
> > > > > __HELO_RECORD needs to be reuse, not PDS_NO_HELO_DNS
> > > > >
> > > > >
> > > > > On Wed, Jul 03, 2019 at 10:04:41AM -0000, [2][3][4]
> pds@apache.org
> > wrote:
> > > > > > Author: pds
> > > > > > Date: Wed Jul 3 10:04:41 2019
> > > > > > New Revision: 1862459
> > > > > >
> > > > > > URL: [3][4][5]http://svn.apache.org/viewvc?rev=1862459&
> view=rev
> > > > > > Log:
> > > > > > These depend on net
> > > > > >
> > > > > > Modified:
> > > > > > spamassassin/trunk/rulesrc/sandbox/pds/[4][5][6]
> 20_helo.cf
> > > > > >
> > > > > > Modified: spamassassin/trunk/rulesrc/sandbox/pds/[5][6]
> > [7]20_helo.cf
> > > > > > URL: [6][7][8]http://svn.apache.org/viewvc/spamassassin/
> trunk/
> > rulesrc/sandbox
> > > > /pds/[8][9]20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=
> diff
> > > > > > =========================================================
> ======
> > ========
> > > > =======
> > > > > > --- spamassassin/trunk/rulesrc/sandbox/pds/[7][9][10]
> 20_helo.cf
> > (original)
> > > > > > +++ spamassassin/trunk/rulesrc/sandbox/pds/[8][10][11]
> 20_helo.cf
> > Wed Jul 3
> > > > 10:04:41 2019
> > > > > > @@ -3,20 +3,24 @@ header __HELO_HIGHPROFILE
> X-Spam-Rela
> > > > > > meta PDS_HELO_SPF_FAIL SPF_HELO_FAIL &&
> __HELO_HIGHPROFILE
> > > > > > describe PDS_HELO_SPF_FAIL High profile HELO that fails
> SPF
> > > > > > score PDS_HELO_SPF_FAIL 2.0
> > > > > > +tflags PDS_HELO_SPF_FAIL net
> > > > > > reuse PDS_HELO_SPF_FAIL
> > > > > >
> > > > > > meta PDS_HP_HELO_NORDNS RDNS_NONE &&
> __HELO_HIGHPROFILE
> > > > > > describe PDS_HP_HELO_NORDNS High profile HELO with no
> sender
> > rDNS
> > > > > > score PDS_HP_HELO_NORDNS 1.0
> > > > > > +tflags PDS_HP_HELO_NORDNS net
> > > > > > reuse PDS_HP_HELO_NORDNS
> > > > > >
> > > > > > ifplugin Mail::SpamAssassin::Plugin::AskDNS
> > > > > >
> > > > > > askdns __HELO_RECORD _LASTEXTERNALHELO_ A /./
> > > > > > +tflags __HELO_RECORD net
> > > > > >
> > > > > > meta PDS_NO_HELO_DNS (__HELO_HIGHPROFILE && !
> > __HELO_RECORD)
> > > > > > describe PDS_NO_HELO_DNS High profile HELO but no A
> record
> > > > > > score PDS_NO_HELO_DNS 2.0
> > > > > > +tflags PDS_NO_HELO_DNS net
> > > > > > reuse PDS_NO_HELO_DNS
> > > > > >
> > > > > > endif
> > > > > >
> > > >
> > > >
> > > > References:
> > > >
> > > > [1] mailto:[11][12]hege@hege.li
> > > > [2] mailto:[12][13]pds@apache.org
> > > > [3] [13][14]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > [4] [14][15]http://20_helo.cf/
> > > > [5] [15][16]http://20_helo.cf/
> > > > [6] [16][17]http://svn.apache.org/viewvc/spamassassin/trunk/
> rulesrc/sandbox
> > /pds/[18]20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > [7] [17][19]http://20_helo.cf/
> > > > [8] [18][20]http://20_helo.cf/
> >
> >
> > References:
> >
> > [1] mailto:[21]hege@hege.li
> > [2] mailto:[22]hege@hege.li
> > [3] mailto:[23]pds@apache.org
> > [4] [24]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > [5] [25]http://20_helo.cf/
> > [6] [26]http://20_helo.cf/
> > [7] [27]http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> > [8] [28]http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> > [9] [29]http://20_helo.cf/
> > [10] [30]http://20_helo.cf/
> > [11] mailto:[31]hege@hege.li
> > [12] mailto:[32]pds@apache.org
> > [13] [33]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > [14] [34]http://20_helo.cf/
> > [15] [35]http://20_helo.cf/
> > [16] [36]http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/
> pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > [17] [37]http://20_helo.cf/
> > [18] [38]http://20_helo.cf/
>
>
> References:
>
> [1] mailto:hege@hege.li
> [2] mailto:hege@hege.li
> [3] mailto:hege@hege.li
> [4] mailto:pds@apache.org
> [5] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [6] http://20_helo.cf/
> [7] http://20_helo.cf/
> [8] http://svn.apache.org/viewvc/spamassassin/trunk/
> [9] http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> [10] http://20_helo.cf/
> [11] http://20_helo.cf/
> [12] mailto:hege@hege.li
> [13] mailto:pds@apache.org
> [14] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [15] http://20_helo.cf/
> [16] http://20_helo.cf/
> [17] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> [18] http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> [19] http://20_helo.cf/
> [20] http://20_helo.cf/
> [21] mailto:hege@hege.li
> [22] mailto:hege@hege.li
> [23] mailto:pds@apache.org
> [24] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [25] http://20_helo.cf/
> [26] http://20_helo.cf/
> [27] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> [28] http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> [29] http://20_helo.cf/
> [30] http://20_helo.cf/
> [31] mailto:hege@hege.li
> [32] mailto:pds@apache.org
> [33] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [34] http://20_helo.cf/
> [35] http://20_helo.cf/
> [36] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> [37] http://20_helo.cf/
> [38] http://20_helo.cf/
Re: svn commit: r1862459 - /spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
Posted by Paul Stead <pa...@gmail.com>.
Mine and other's initial scoring is done within an instance of Amavis on
regular mail servers then masschecked with trunk - any adjustments would
have to be with the ruleqa/masscheck code on sa-vm1
Paul
On Wed, 3 Jul 2019 at 11:53, Henrik K <he...@hege.li> wrote:
>
> Since reuse is pretty much only used for masscheckers who use trunk, we
> should be able to figure out something..
>
> net tflag is really only used in Check.pm for skipping rules in non-net
> scoreset and waiting for net rule dependencies to finish. Perhaps it would
> be safe to inherit net flag from meta subrules automatically (basically
> this
> is what is already done manually).
>
>
> On Wed, Jul 03, 2019 at 11:32:35AM +0100, Paul Stead wrote:
> > +1 to the crazy bit, it's something I've realised over the weeks of
> looking
> > into the mass check.
> >
> > With reuse flag on masscheck the X-Spam headers are used to generate the
> corpus
> > submission - subrules aren't output in this header so they're not
> submitted.
> >
> > I'm not certain how to fix without requiring full rule list in headers -
> this
> > causes problems across multiple SA instances and versions - Amavis etc
> >
> > On Wed, 3 Jul 2019 at 11:30, Henrik Krohns <[1...@hege.li> wrote:
> >
> >
> > There's also some strange dependency for autolearning in Conf.pm
> functions
> > maybe_header_only, maybe_body_only (who came up with the function
> names?
> > so
> > vague). It seems net flagged metas are ignored for auto-learn
> scoring.
> > Dunno if this is good or bad, more stuff to investigate..
> >
> > Cc'ing dev
> >
> > On Wed, Jul 03, 2019 at 01:26:11PM +0300, Henrik Krohns wrote:
> > >
> > > That should be considered a bug then.
> > >
> > > Probably metas should inherit all the flags from rules they use,
> it's
> > crazy
> > > to start flagging all meta rules manually.
> > >
> > > On Wed, Jul 03, 2019 at 11:21:02AM +0100, Paul Stead wrote:
> > > > On the weekly mass check when --reuse is used subrules are not
> > submitted. This
> > > > means if no reuse on the full rule it'll be ignored during a
> --reuse
> > run
> > > > because the subrule isn't submitted.
> > > >
> > > > On Wed, 3 Jul 2019 at 11:10, Henrik Krohns <[1...@hege.li>
> wrote:
> > > >
> > > >
> > > > also no need to reuse PDS_HP_HELO_NORDNS, metas do not launch
> > network
> > > > rules.
> > > > Reuse needs to be on the actual query launching rule.
> > > >
> > > > On Wed, Jul 03, 2019 at 01:09:19PM +0300, Henrik K wrote:
> > > > >
> > > > >
> > > > > __HELO_RECORD needs to be reuse, not PDS_NO_HELO_DNS
> > > > >
> > > > >
> > > > > On Wed, Jul 03, 2019 at 10:04:41AM -0000, [2][3]
> pds@apache.org
> > wrote:
> > > > > > Author: pds
> > > > > > Date: Wed Jul 3 10:04:41 2019
> > > > > > New Revision: 1862459
> > > > > >
> > > > > > URL: [3][4]
> http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > > > Log:
> > > > > > These depend on net
> > > > > >
> > > > > > Modified:
> > > > > > spamassassin/trunk/rulesrc/sandbox/pds/[4][5]
> 20_helo.cf
> > > > > >
> > > > > > Modified: spamassassin/trunk/rulesrc/sandbox/pds/[5][6]
> > 20_helo.cf
> > > > > > URL: [6][7]
> http://svn.apache.org/viewvc/spamassassin/trunk/
> > rulesrc/sandbox
> > > > /pds/[8]
> 20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > > >
> ===============================================================
> > ========
> > > > =======
> > > > > > --- spamassassin/trunk/rulesrc/sandbox/pds/[7][9]
> 20_helo.cf
> > (original)
> > > > > > +++ spamassassin/trunk/rulesrc/sandbox/pds/[8][10]
> 20_helo.cf
> > Wed Jul 3
> > > > 10:04:41 2019
> > > > > > @@ -3,20 +3,24 @@ header __HELO_HIGHPROFILE
> X-Spam-Rela
> > > > > > meta PDS_HELO_SPF_FAIL SPF_HELO_FAIL &&
> __HELO_HIGHPROFILE
> > > > > > describe PDS_HELO_SPF_FAIL High profile HELO that fails
> SPF
> > > > > > score PDS_HELO_SPF_FAIL 2.0
> > > > > > +tflags PDS_HELO_SPF_FAIL net
> > > > > > reuse PDS_HELO_SPF_FAIL
> > > > > >
> > > > > > meta PDS_HP_HELO_NORDNS RDNS_NONE &&
> __HELO_HIGHPROFILE
> > > > > > describe PDS_HP_HELO_NORDNS High profile HELO with no
> sender
> > rDNS
> > > > > > score PDS_HP_HELO_NORDNS 1.0
> > > > > > +tflags PDS_HP_HELO_NORDNS net
> > > > > > reuse PDS_HP_HELO_NORDNS
> > > > > >
> > > > > > ifplugin Mail::SpamAssassin::Plugin::AskDNS
> > > > > >
> > > > > > askdns __HELO_RECORD _LASTEXTERNALHELO_ A /./
> > > > > > +tflags __HELO_RECORD net
> > > > > >
> > > > > > meta PDS_NO_HELO_DNS (__HELO_HIGHPROFILE && !
> > __HELO_RECORD)
> > > > > > describe PDS_NO_HELO_DNS High profile HELO but no A
> record
> > > > > > score PDS_NO_HELO_DNS 2.0
> > > > > > +tflags PDS_NO_HELO_DNS net
> > > > > > reuse PDS_NO_HELO_DNS
> > > > > >
> > > > > > endif
> > > > > >
> > > >
> > > >
> > > > References:
> > > >
> > > > [1] mailto:[11]hege@hege.li
> > > > [2] mailto:[12]pds@apache.org
> > > > [3] [13]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > [4] [14]http://20_helo.cf/
> > > > [5] [15]http://20_helo.cf/
> > > > [6] [16]
> http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> > /pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > [7] [17]http://20_helo.cf/
> > > > [8] [18]http://20_helo.cf/
> >
> >
> > References:
> >
> > [1] mailto:hege@hege.li
> > [2] mailto:hege@hege.li
> > [3] mailto:pds@apache.org
> > [4] http://svn.apache.org/viewvc?rev=1862459&view=rev
> > [5] http://20_helo.cf/
> > [6] http://20_helo.cf/
> > [7] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> > [8] http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> > [9] http://20_helo.cf/
> > [10] http://20_helo.cf/
> > [11] mailto:hege@hege.li
> > [12] mailto:pds@apache.org
> > [13] http://svn.apache.org/viewvc?rev=1862459&view=rev
> > [14] http://20_helo.cf/
> > [15] http://20_helo.cf/
> > [16]
> http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > [17] http://20_helo.cf/
> > [18] http://20_helo.cf/
>
Re: svn commit: r1862459 -
/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
Posted by Henrik K <he...@hege.li>.
Since reuse is pretty much only used for masscheckers who use trunk, we
should be able to figure out something..
net tflag is really only used in Check.pm for skipping rules in non-net
scoreset and waiting for net rule dependencies to finish. Perhaps it would
be safe to inherit net flag from meta subrules automatically (basically this
is what is already done manually).
On Wed, Jul 03, 2019 at 11:32:35AM +0100, Paul Stead wrote:
> +1 to the crazy bit, it's something I've realised over the weeks of looking
> into the mass check.
>
> With reuse flag on masscheck the X-Spam headers are used to generate the corpus
> submission - subrules aren't output in this header so they're not submitted.
>
> I'm not certain how to fix without requiring full rule list in headers - this
> causes problems across multiple SA instances and versions - Amavis etc
>
> On Wed, 3 Jul 2019 at 11:30, Henrik Krohns <[1...@hege.li> wrote:
>
>
> There's also some strange dependency for autolearning in Conf.pm functions
> maybe_header_only, maybe_body_only (who came up with the function names?
> so
> vague). It seems net flagged metas are ignored for auto-learn scoring.
> Dunno if this is good or bad, more stuff to investigate..
>
> Cc'ing dev
>
> On Wed, Jul 03, 2019 at 01:26:11PM +0300, Henrik Krohns wrote:
> >
> > That should be considered a bug then.
> >
> > Probably metas should inherit all the flags from rules they use, it's
> crazy
> > to start flagging all meta rules manually.
> >
> > On Wed, Jul 03, 2019 at 11:21:02AM +0100, Paul Stead wrote:
> > > On the weekly mass check when --reuse is used subrules are not
> submitted. This
> > > means if no reuse on the full rule it'll be ignored during a --reuse
> run
> > > because the subrule isn't submitted.
> > >
> > > On Wed, 3 Jul 2019 at 11:10, Henrik Krohns <[1...@hege.li> wrote:
> > >
> > >
> > > also no need to reuse PDS_HP_HELO_NORDNS, metas do not launch
> network
> > > rules.
> > > Reuse needs to be on the actual query launching rule.
> > >
> > > On Wed, Jul 03, 2019 at 01:09:19PM +0300, Henrik K wrote:
> > > >
> > > >
> > > > __HELO_RECORD needs to be reuse, not PDS_NO_HELO_DNS
> > > >
> > > >
> > > > On Wed, Jul 03, 2019 at 10:04:41AM -0000, [2][3]pds@apache.org
> wrote:
> > > > > Author: pds
> > > > > Date: Wed Jul 3 10:04:41 2019
> > > > > New Revision: 1862459
> > > > >
> > > > > URL: [3][4]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > > Log:
> > > > > These depend on net
> > > > >
> > > > > Modified:
> > > > > spamassassin/trunk/rulesrc/sandbox/pds/[4][5]20_helo.cf
> > > > >
> > > > > Modified: spamassassin/trunk/rulesrc/sandbox/pds/[5][6]
> 20_helo.cf
> > > > > URL: [6][7]http://svn.apache.org/viewvc/spamassassin/trunk/
> rulesrc/sandbox
> > > /pds/[8]20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > > ===============================================================
> ========
> > > =======
> > > > > --- spamassassin/trunk/rulesrc/sandbox/pds/[7][9]20_helo.cf
> (original)
> > > > > +++ spamassassin/trunk/rulesrc/sandbox/pds/[8][10]20_helo.cf
> Wed Jul 3
> > > 10:04:41 2019
> > > > > @@ -3,20 +3,24 @@ header __HELO_HIGHPROFILE X-Spam-Rela
> > > > > meta PDS_HELO_SPF_FAIL SPF_HELO_FAIL && __HELO_HIGHPROFILE
> > > > > describe PDS_HELO_SPF_FAIL High profile HELO that fails SPF
> > > > > score PDS_HELO_SPF_FAIL 2.0
> > > > > +tflags PDS_HELO_SPF_FAIL net
> > > > > reuse PDS_HELO_SPF_FAIL
> > > > >
> > > > > meta PDS_HP_HELO_NORDNS RDNS_NONE && __HELO_HIGHPROFILE
> > > > > describe PDS_HP_HELO_NORDNS High profile HELO with no sender
> rDNS
> > > > > score PDS_HP_HELO_NORDNS 1.0
> > > > > +tflags PDS_HP_HELO_NORDNS net
> > > > > reuse PDS_HP_HELO_NORDNS
> > > > >
> > > > > ifplugin Mail::SpamAssassin::Plugin::AskDNS
> > > > >
> > > > > askdns __HELO_RECORD _LASTEXTERNALHELO_ A /./
> > > > > +tflags __HELO_RECORD net
> > > > >
> > > > > meta PDS_NO_HELO_DNS (__HELO_HIGHPROFILE && !
> __HELO_RECORD)
> > > > > describe PDS_NO_HELO_DNS High profile HELO but no A record
> > > > > score PDS_NO_HELO_DNS 2.0
> > > > > +tflags PDS_NO_HELO_DNS net
> > > > > reuse PDS_NO_HELO_DNS
> > > > >
> > > > > endif
> > > > >
> > >
> > >
> > > References:
> > >
> > > [1] mailto:[11]hege@hege.li
> > > [2] mailto:[12]pds@apache.org
> > > [3] [13]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > [4] [14]http://20_helo.cf/
> > > [5] [15]http://20_helo.cf/
> > > [6] [16]http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> /pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > [7] [17]http://20_helo.cf/
> > > [8] [18]http://20_helo.cf/
>
>
> References:
>
> [1] mailto:hege@hege.li
> [2] mailto:hege@hege.li
> [3] mailto:pds@apache.org
> [4] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [5] http://20_helo.cf/
> [6] http://20_helo.cf/
> [7] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> [8] http://20_helo.cf/?rev=1862459&r1=1862458&r2=1862459&view=diff
> [9] http://20_helo.cf/
> [10] http://20_helo.cf/
> [11] mailto:hege@hege.li
> [12] mailto:pds@apache.org
> [13] http://svn.apache.org/viewvc?rev=1862459&view=rev
> [14] http://20_helo.cf/
> [15] http://20_helo.cf/
> [16] http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> [17] http://20_helo.cf/
> [18] http://20_helo.cf/
Re: svn commit: r1862459 - /spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf
Posted by Paul Stead <pa...@gmail.com>.
+1 to the crazy bit, it's something I've realised over the weeks of looking
into the mass check.
With reuse flag on masscheck the X-Spam headers are used to generate the
corpus submission - subrules aren't output in this header so they're not
submitted.
I'm not certain how to fix without requiring full rule list in headers -
this causes problems across multiple SA instances and versions - Amavis etc
On Wed, 3 Jul 2019 at 11:30, Henrik Krohns <he...@hege.li> wrote:
>
> There's also some strange dependency for autolearning in Conf.pm functions
> maybe_header_only, maybe_body_only (who came up with the function names?
> so
> vague). It seems net flagged metas are ignored for auto-learn scoring.
> Dunno if this is good or bad, more stuff to investigate..
>
> Cc'ing dev
>
> On Wed, Jul 03, 2019 at 01:26:11PM +0300, Henrik Krohns wrote:
> >
> > That should be considered a bug then.
> >
> > Probably metas should inherit all the flags from rules they use, it's
> crazy
> > to start flagging all meta rules manually.
> >
> > On Wed, Jul 03, 2019 at 11:21:02AM +0100, Paul Stead wrote:
> > > On the weekly mass check when --reuse is used subrules are not
> submitted. This
> > > means if no reuse on the full rule it'll be ignored during a --reuse
> run
> > > because the subrule isn't submitted.
> > >
> > > On Wed, 3 Jul 2019 at 11:10, Henrik Krohns <[1...@hege.li> wrote:
> > >
> > >
> > > also no need to reuse PDS_HP_HELO_NORDNS, metas do not launch
> network
> > > rules.
> > > Reuse needs to be on the actual query launching rule.
> > >
> > > On Wed, Jul 03, 2019 at 01:09:19PM +0300, Henrik K wrote:
> > > >
> > > >
> > > > __HELO_RECORD needs to be reuse, not PDS_NO_HELO_DNS
> > > >
> > > >
> > > > On Wed, Jul 03, 2019 at 10:04:41AM -0000, [2]pds@apache.org
> wrote:
> > > > > Author: pds
> > > > > Date: Wed Jul 3 10:04:41 2019
> > > > > New Revision: 1862459
> > > > >
> > > > > URL: [3]http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > > > Log:
> > > > > These depend on net
> > > > >
> > > > > Modified:
> > > > > spamassassin/trunk/rulesrc/sandbox/pds/[4]20_helo.cf
> > > > >
> > > > > Modified: spamassassin/trunk/rulesrc/sandbox/pds/[5]20_helo.cf
> > > > > URL: [6]
> http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox
> > > /pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > > >
> =======================================================================
> > > =======
> > > > > --- spamassassin/trunk/rulesrc/sandbox/pds/[7]20_helo.cf
> (original)
> > > > > +++ spamassassin/trunk/rulesrc/sandbox/pds/[8]20_helo.cf Wed
> Jul 3
> > > 10:04:41 2019
> > > > > @@ -3,20 +3,24 @@ header __HELO_HIGHPROFILE X-Spam-Rela
> > > > > meta PDS_HELO_SPF_FAIL SPF_HELO_FAIL && __HELO_HIGHPROFILE
> > > > > describe PDS_HELO_SPF_FAIL High profile HELO that fails SPF
> > > > > score PDS_HELO_SPF_FAIL 2.0
> > > > > +tflags PDS_HELO_SPF_FAIL net
> > > > > reuse PDS_HELO_SPF_FAIL
> > > > >
> > > > > meta PDS_HP_HELO_NORDNS RDNS_NONE && __HELO_HIGHPROFILE
> > > > > describe PDS_HP_HELO_NORDNS High profile HELO with no sender
> rDNS
> > > > > score PDS_HP_HELO_NORDNS 1.0
> > > > > +tflags PDS_HP_HELO_NORDNS net
> > > > > reuse PDS_HP_HELO_NORDNS
> > > > >
> > > > > ifplugin Mail::SpamAssassin::Plugin::AskDNS
> > > > >
> > > > > askdns __HELO_RECORD _LASTEXTERNALHELO_ A /./
> > > > > +tflags __HELO_RECORD net
> > > > >
> > > > > meta PDS_NO_HELO_DNS (__HELO_HIGHPROFILE &&
> !__HELO_RECORD)
> > > > > describe PDS_NO_HELO_DNS High profile HELO but no A record
> > > > > score PDS_NO_HELO_DNS 2.0
> > > > > +tflags PDS_NO_HELO_DNS net
> > > > > reuse PDS_NO_HELO_DNS
> > > > >
> > > > > endif
> > > > >
> > >
> > >
> > > References:
> > >
> > > [1] mailto:hege@hege.li
> > > [2] mailto:pds@apache.org
> > > [3] http://svn.apache.org/viewvc?rev=1862459&view=rev
> > > [4] http://20_helo.cf/
> > > [5] http://20_helo.cf/
> > > [6]
> http://svn.apache.org/viewvc/spamassassin/trunk/rulesrc/sandbox/pds/20_helo.cf?rev=1862459&r1=1862458&r2=1862459&view=diff
> > > [7] http://20_helo.cf/
> > > [8] http://20_helo.cf/
>