Policy on cross-posting to multiple lists

Dean Rasheed <dean.a.rasheed@gmail.com> writes:

Has the policy on cross-posting to multiple lists been hardened recently?

Not that I've heard of.

The "Crash on ALTER TABLE" thread [1] started on -bugs, but Andrew's
message on 8 Jan with an initial proposed patch and my response later
that day both CC'ed -hackers and seem to have been rejected, and so
are missing from the archives.

I've done the same recently, without problems. I'd suggest inquiring
on the pgsql-www list; project infrastructure issues are not really
on-topic here.

regards, tom lane

On Fri, Jan 11, 2019 at 12:58 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:

Dean Rasheed <dean.a.rasheed@gmail.com> writes:

The "Crash on ALTER TABLE" thread [1] started on -bugs, but Andrew's
message on 8 Jan with an initial proposed patch and my response later
that day both CC'ed -hackers and seem to have been rejected, and so
are missing from the archives.

I've done the same recently, without problems. I'd suggest inquiring
on the pgsql-www list; project infrastructure issues are not really
on-topic here.

Fwiw, an email I sent yesterday was rejected similarly because I'd
tried to send it to both pgsql-hackers and pgsql-performance. I
mentioned about that when I resent the same email successfully [1]/messages/by-id/96720c99-ffa0-01ad-c594-0504c8eda708@lab.ntt.co.jp
after dropping pgsql-performance from the list of recipients.

Thanks,
Amit

[1]: /messages/by-id/96720c99-ffa0-01ad-c594-0504c8eda708@lab.ntt.co.jp

Greetings,

(moving to -www as suggested downthread and as generally more
appropriate)

* Dean Rasheed (dean.a.rasheed@gmail.com) wrote:

Has the policy on cross-posting to multiple lists been hardened recently?

So, the short answer is 'yes'.

We've made a few different changes in the recent weeks. The first
change that was made was actually to start dropping emails where the
list is being BCC'd. That was done a couple of weeks ago and seems to
have gone well and has reduced the amount of spam our moderators are
dealing with.

This most recent change was to implement a policy where we don't allow
public lists to be CC'd with other public lists; when that happens we
instead reply with an email basically saying "please pick the right list
to send your email to."

Perhaps that hasn't been getting through to people...? Though I had
someone respond to -owner basically saying "thanks, I'll pick the right
list", so at least some are seeing it.

As for how this change came to be implemented without much discussion
externally, I'm afraid that's probably the combination of "well, the BCC
change went just fine and no one complained", confusion between folks on
infra as to if we had only discussed it internally or if we had already
discussed it externally with people (the individual who actually made
the change *cough* apparently thought it had already been discussed
externally when we hadn't and probably should have at least announced
it when we did make the change anyway...), and general frustration among
some about the increasing number of cross-post emails we're getting
which really shouldn't be cross-posted.

In an ideal world, everyone would know that they really *shouldn't*
cross-post, and we also wouldn't have extremely long many-mailing-list
cross-posted threads, and we wouldn't need to have such a policy, but
that's not really where we are.

One thing which hadn't been considered and probably should have is the
impact on existing threads, but I'm not sure if we really could have
sensibly done something about that.

Then there's the big question which we really should have discussed
ahead of time, but, do people feel that such a restriction ends up doing
more harm than good? Are there concerns about the BCC restriction? In
the short period of time that it's been in place, I've seen some good
come from it in the form of people learning to post to the correct list
instead of just cross-posting to a bunch of lists, but I've also seen
(now) the cases where existing threads were confused do to the change,
so I suppose I'm on the fence, though I still tend towards having the
policy in place and hoping that it doesn't overly bother existing users
while helping newcomers.

We're here now though, so, thoughts? Should I go undo it right away?
Should we see how it goes? Try other things? We could possibly have it
only apply to emails from people who don't have accounts or who aren't
subscribed to the lists? Or have a flag on a per-account basis which
basically says "let me cross-post"? Open to suggestions (note: I've not
run all the above ideas by the other pglister hacker *cough*, so I can't
say if all of them would be possible/reasonable, just throwing out
ideas).

Thanks!

Stephen

On 2019-Jan-10, Stephen Frost wrote:

Are there concerns about the BCC restriction?

None here.

We're here now though, so, thoughts? Should I go undo it right away?

I don't like the crosspost ban, personally. Some sort of limit makes
sense, but I think cross-posting to two lists should be allowed. I
don't see an use case for cross-posting to more than two lists (though
maybe -hackers + -bugs + -docs would make sense ...)

--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Stephen Frost <sfrost@snowman.net> writes:

* Dean Rasheed (dean.a.rasheed@gmail.com) wrote:

Has the policy on cross-posting to multiple lists been hardened recently?

So, the short answer is 'yes'.
Perhaps that hasn't been getting through to people...?

If this was publicly announced anywhere, I didn't see it.
I would have pushed back if I had. CC'ing -hackers on a reply to
a bug report is something I do all the time, and I do not think
it'd be a good idea to stop doing so, nor to make the thread
disappear from the -bugs archives.

I'm quite on board with the need to reduce useless cross-posting,
but this is not the solution.

Maybe there could be a different rule for initial submissions
(one list only) than follow-ups (can add lists)?

regards, tom lane

On Thu, 10 Jan 2019 at 17:18, Stephen Frost <sfrost@snowman.net> wrote:

* Dean Rasheed (dean.a.rasheed@gmail.com) wrote:

Has the policy on cross-posting to multiple lists been hardened recently?

This most recent change was to implement a policy where we don't allow
public lists to be CC'd with other public lists; when that happens we
instead reply with an email basically saying "please pick the right list
to send your email to."

The problem with that as a mechanism for stopping people from cross
posting is that it doesn't (and can't) actually stop the message from
being delivered to people already on the CC list for that thread.

So in this case, Andrew first cross posted it, but I was already on
the CC list, so I got the message as normal, not realising that it
hadn't come via the lists. I then hit "Reply all" ... (rinse and
repeat). I didn't even immediately notice the failure to send the
message because my own reply just got added to end of the conversation
in my mail client, but presumably the intention was that both Andrew
and I should have noticed and re-posted to a single list. But of
course that would then have annoyed all the people already on the
thread who would have got duplicates of mails they had already
received.

Personally, I don't have a problem with people cross posting. I think
there are real cases where it's the right thing to do -- it's common
practice for legitimate reasons. Yes, it can be abused, but there are
worse abuses of email all the time.

Regards,
Dean

On Thu, Jan 10, 2019 at 11:17 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:

Stephen Frost <sfrost@snowman.net> writes:

* Dean Rasheed (dean.a.rasheed@gmail.com) wrote:

Has the policy on cross-posting to multiple lists been hardened recently?

So, the short answer is 'yes'.
Perhaps that hasn't been getting through to people...?

If this was publicly announced anywhere, I didn't see it.
I would have pushed back if I had. CC'ing -hackers on a reply to
a bug report is something I do all the time, and I do not think
it'd be a good idea to stop doing so, nor to make the thread
disappear from the -bugs archives.

I'm quite on board with the need to reduce useless cross-posting,
but this is not the solution.

Agreed. Similarly, posts from pgadmin-support sometimes end up
intentionally being cross posted to pgadmin-hackers.

--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Greetings,

* Dave Page (dpage@pgadmin.org) wrote:

On Thu, Jan 10, 2019 at 11:17 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:

I'm quite on board with the need to reduce useless cross-posting,
but this is not the solution.

Agreed. Similarly, posts from pgadmin-support sometimes end up
intentionally being cross posted to pgadmin-hackers.

So, in implementing this we did consider that different lists might wish
for different policies and given that -hackers seems to be common among
the discussion, what if we just dropped the restriction for posts to
-hackers?

That is, emails to -bugs and -hackers would be allowed through to both
lists, cross-posts to -general and -sql, for example, would get the
bounce-back.

As there seems relatively little downside, I've gone ahead and made that
change, but I don't mean to forstall further discussion. Should we
apply that change to other lists? To all of them?

Tom's idea about allowing cross-posts on replies is an interesting one
as well. I've also added a certain someone to the thread explicitly to
see what his thoughts are on that, and the rest.

Thanks!

Stephen

On Thu, Jan 10, 2019 at 12:18 PM Stephen Frost <sfrost@snowman.net> wrote:

Are there concerns about the BCC restriction?

One thing people sometimes do when something is posted to the wrong
list is (1) reply, (2) explain in the reply that the message was
posted to the wrong pace, (3) move the original list from Cc into Bcc,
and (4) add the correct list into Cc. That has the advantage that
people on the original list can see that someone replied (which avoids
duplicate replies by different people) and know where to go to find
the rest of the discussion if they want to see it.

I think the idea of allowing 2 lists but not >2 is probably a good
one. Also, it might be good to be more permissive for, say, people
who have successfully posted at least 1000 emails to the lists. Such
people presumably are less likely to do abusive things, and more
likely to care about and heed any correction given to them.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Greetings,

* Robert Haas (robertmhaas@gmail.com) wrote:

On Thu, Jan 10, 2019 at 12:18 PM Stephen Frost <sfrost@snowman.net> wrote:

Are there concerns about the BCC restriction?

One thing people sometimes do when something is posted to the wrong
list is (1) reply, (2) explain in the reply that the message was
posted to the wrong pace, (3) move the original list from Cc into Bcc,
and (4) add the correct list into Cc. That has the advantage that
people on the original list can see that someone replied (which avoids
duplicate replies by different people) and know where to go to find
the rest of the discussion if they want to see it.

While considering that, we actively went and looked at both the
frequency and the success of that approach and, frankly, neither were
very inspiring. There were very few cases of that being tried and, as I
recall anyway, none of them were actually successful in 'moving' the
thread- that is, people continued on the original list to begin with
anyway, except that some of the thread was now on another list.

We had discussed allowing bcc's to lists when we detect that there's at
least *some* valid list in the To or Cc line, but it didn't seem
worthwhile given the research that was done.

For some (private) lists, we have the policy set to moderate emails
which bcc those lists (such as -security). Also, the "don't CC multiple
lists" was only applied to public/archived lists to begin with,
intentionally.

I think the idea of allowing 2 lists but not >2 is probably a good
one. Also, it might be good to be more permissive for, say, people
who have successfully posted at least 1000 emails to the lists. Such
people presumably are less likely to do abusive things, and more
likely to care about and heed any correction given to them.

Yeah, that's in-line with what I had suggested up-thread where we have
some kind of flag which can either be set by the user themselves (maybe
we have some language above the flag that cautions against cross-posts
and whatnot), or set by the system (>1000 emails, as you say, or maybe
"after 2 weeks of being subscribed to a list", similar to the community
account "cooling off" period we have), or maybe by the list admins
(likely initially based on a heuristic of "lots of emails sent" or
something, but then handled on an individual basis).

I am a little concerned that we make the system too complicated for
people to understand too though. Haven't got a particularly good answer
for that, sadly.

Thanks!

Stephen

Hi,

On 2019-01-10 12:47:03 -0500, Tom Lane wrote:

Stephen Frost <sfrost@snowman.net> writes:

* Dean Rasheed (dean.a.rasheed@gmail.com) wrote:

Has the policy on cross-posting to multiple lists been hardened recently?

So, the short answer is 'yes'.
Perhaps that hasn't been getting through to people...?

If this was publicly announced anywhere, I didn't see it.
I would have pushed back if I had. CC'ing -hackers on a reply to
a bug report is something I do all the time, and I do not think
it'd be a good idea to stop doing so, nor to make the thread
disappear from the -bugs archives.

+1

This seems quite the significant change to make without public
discussion.

Greetings,

Andres Freund

On 2019-01-10 12:18:35 -0500, Stephen Frost wrote:

Should I go undo it right away?

Yes.

I'm quite on board with the need to reduce useless cross-posting,
but this is not the solution.

Agreed. Similarly, posts from pgadmin-support sometimes end up
intentionally being cross posted to pgadmin-hackers.

Another use case is, pgsql-docs and pgsql-hackers. For non trivial
documentation changes I would like to register a doc patch to CF, but
CF app does not pick up any messages other than posted in
pgsql-hackers. So to discuss with pgsql-doc subscribers, while dealing
with CF app, I would like cross postings for pgsql-hackers and
pgsql-docs.

Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp

Greetings,

* Tatsuo Ishii (ishii@sraoss.co.jp) wrote:

I'm quite on board with the need to reduce useless cross-posting,
but this is not the solution.

Agreed. Similarly, posts from pgadmin-support sometimes end up
intentionally being cross posted to pgadmin-hackers.

Another use case is, pgsql-docs and pgsql-hackers. For non trivial
documentation changes I would like to register a doc patch to CF, but
CF app does not pick up any messages other than posted in
pgsql-hackers. So to discuss with pgsql-doc subscribers, while dealing
with CF app, I would like cross postings for pgsql-hackers and
pgsql-docs.

So, cross-posting between -hackers and -docs should be working now,
thanks to the change I made yesterday.

After stealing some time from Magnus to chat quickly about this (he
seems to be mostly unavailable at present), what we're trying to figure
out is what the group, overall, wants, and in particular if the change
to allow cross-posting with -hackers solves the valid use-cases while
preventing the invalid use-cases (like cross-posting between -general,
-performance, and -sql).

Of course, it isn't perfect, but then it's unlikely that anything will
be. Changes which require us to write additional code into pglister
will, of course, take longer, but we can work towards it if there's
agreement about what such a change would look like. In the interim, we
could see how things go with the current configuration, or we could add
other lists to the 'exclude', beyond just -hackers and the private
lists, or we could add them all (effectively going back to where things
were before the changes were made).

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

Thanks!

Stephen

On Fri, Jan 11, 2019 at 11:38 AM Stephen Frost <sfrost@snowman.net> wrote:

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

Well, you know, you could just undo the ban which you imposed
unilaterally and which nobody so far has said they liked, and multiple
people have said they disliked. Then after having the public
discussion about what the policy should be, you could implement the
conclusions of that discussion.

I mean, personally, I have no problem with SOME cross-posting
restrictions, but nothing you've proposed so far seems very good,
other than maybe the >2 rule. But if you're looking to understand
what people want better, you don't really need more votes. What has
been said by a whole bunch of people is not in any significant way
unclear. They don't like the restrictions, and they do like being
consulted.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Hi,

On 2019-01-11 11:38:05 -0500, Stephen Frost wrote:

After stealing some time from Magnus to chat quickly about this (he
seems to be mostly unavailable at present), what we're trying to figure
out is what the group, overall, wants, and in particular if the change
to allow cross-posting with -hackers solves the valid use-cases while
preventing the invalid use-cases (like cross-posting between -general,
-performance, and -sql).

Those don't really seem to be common and painful enough to really need a
technical solution. -performance still seems like a useful subset of
people, and sometimes threads migrate to/from there. I'd personally just
merge -sql with -general, it doesn't seem to have a use-case left
anymore. But that can be done later.

Of course, it isn't perfect, but then it's unlikely that anything will
be. Changes which require us to write additional code into pglister
will, of course, take longer, but we can work towards it if there's
agreement about what such a change would look like. In the interim, we
could see how things go with the current configuration, or we could add
other lists to the 'exclude', beyond just -hackers and the private
lists, or we could add them all (effectively going back to where things
were before the changes were made).

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

I think you should just revert to the prior state, and then we can
discuss potential solutions and the problems they're intended to
address. I find it baffling that after being called out for
unilateral/not publicly discussed decisions you attempt to address that
criticism by continuing to make unilateral decisions.

Greetings,

Andres Freund

On 2019-Jan-11, Andres Freund wrote:

On 2019-01-11 11:38:05 -0500, Stephen Frost wrote:

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

I think you should just revert to the prior state, and then we can
discuss potential solutions and the problems they're intended to
address.

+1

--
�lvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

On Fri, Jan 11, 2019 at 7:33 PM Alvaro Herrera <alvherre@2ndquadrant.com>
wrote:

On 2019-Jan-11, Andres Freund wrote:

On 2019-01-11 11:38:05 -0500, Stephen Frost wrote:

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

I think you should just revert to the prior state, and then we can
discuss potential solutions and the problems they're intended to
address.

+1

I've reverted this change across all lists it was enabled for.

And for the record, I'm the one who asked Stephen to go for a second round
of feedback and not just immediately revert it (he pinged me on chat, as I
was unable to keep track of the mail thread myself due to other commitments
and airplanes and things).

--
Magnus Hagander
Me: https://www.hagander.net/ <http://www.hagander.net/&gt;
Work: https://www.redpill-linpro.com/ <http://www.redpill-linpro.com/&gt;

Alvaro Herrera <alvherre@2ndquadrant.com> writes:

On 2019-Jan-11, Andres Freund wrote:

On 2019-01-11 11:38:05 -0500, Stephen Frost wrote:

Thoughts? Specific votes in one of those directions would help me, at
least, figure out what should be done today.

I think you should just revert to the prior state, and then we can
discuss potential solutions and the problems they're intended to
address.

+1

Same here. The problem you want to solve has been there for decades,
we don't need a solution urgently.

regards, tom lane