You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@buildstream.apache.org by Tristan Van Berkom <tr...@codethink.co.uk> on 2020/07/07 09:26:34 UTC

Can we please disable harmful "Reply-to" munging ?

Dear mailing list admins,

I've noticed a strange behavior while replying to emails on this new
list, as if the list is trying to control to whom I send the replies,
and upon inspection, I have verified that this list is indeed munging
the reply-to headers.

As this has been well known to be harmful for decades[0], I would ask,
can we please disable this reply-to munging on our mailing list ?

Especially for the sake of following the principles of least surprise
and the least damage, as specified in the (very old) referred article,
the inconvenience of occasionally receiving multiple copies of the same
email (one to you, and one to the list, and only when using "reply all"
instead of "reply to list" which some modern mail clients also now
support), is a very small price to pay.

Best regards,
    -Tristan

[0]: https://www.unicom.com/pw/reply-to-harmful.html



Re: Can we please disable harmful "Reply-to" munging ?

Posted by Sander Striker <st...@apache.org>.
Hi Daniel,

Thanks for posting.

On Fri, Jul 31, 2020, 09:16 Daniel Shahaf <d....@daniel.shahaf.name> wrote:

> Tristan Van Berkom wrote on Wed, Jul 08, 2020 at 23:00:36 +0900:
> > On Tue, 2020-07-07 at 21:34 +0200, Sander Striker wrote:
> > > Given the above, I'm inclined to just leave it as-is.  But obviously we
> > > could discuss further if need be.
> >
> > While I certainly don't agree with it, it would appear that the ASF has
> > already taken a project-wide position on this,
>
> The ASF does not have a cross-project position on Reply-To adding.
>

You are correct on that.


> The ASF exists to enable projects to do their thing.


To put this a bit more in perspective I'll quote the ASF mission
statement: The mission of the Apache Software Foundation (ASF) is to
provide software for the public good. We do this by providing services and
support for many like-minded software project communities consisting of
individuals who choose to participate in ASF activities.


> Among other things, that means the ASF doesn't tell projects how to run
> their
> mailing lists.


We provide infrastructure for projects, and allow for customization within
reason.  This ranges from websites, to email, to version control, etc etc.


>   Project lists are created with Reply-To adding enabled
> by default, but that's just that: a default, not a mandate.  It is
> dev@foo.apache.org's decision whether or not dev@foo.apache.org should
> set Reply-To.
>
> (Personally, I would have expected the Apache lists to be created by
> default with the same Reply-To settings as the respective pre-Apache
> lists.  There used to be a checkbox for this in the mailing list
> creation request form.)
>

I think the previous mailinglist was created with defaults common to Gnome,
and here we simply did the same.  Honestly I for one didn't expect any
controversy over this setting, and didn't give it any thought prior this
thread.


> ---
>
> Another point of view: ASF doesn't choose winners.  That's why ASF is
> happy to host competing projects and is allergic to majority votes.


I'd rather state this as: we prefer to get to decisions by reaching
consensus, relying on votes as a last resort.

  Any
> sort of "Project dev@ lists MUST be configured to do X" policy would be
> a choice of winner, and so would need very good reasons.
>
> ---
>
> Not directly related, but there are at least three options for y'all to
> choose from:
>
> - Add no headers.
> - Add Reply-To.
> - Add Mail-Followup-To.
>
> > > I remember from my early days in the ASF (2004) that we also had a
> short
> > > conversation about Reply-To [4].  The expectation seemed to be that
> replies
> > > went to the list.  And annoyance was voiced more regarding
> inconsistency
> > > than anything else.  The result was the inconsistency being fixed.
> > >
>
> So, let's formalize your argument for a minute, Sander:
>
> Premise: The Apache HTTP Server developers decided in 2004 to configure
> their dev@ list in the same way their spun-off sister project's dev@
> list was configured.
>
> Conclusion: Buildstream should configure its lists the same way today.
>
> I don't see how the conclusion follows from the premise.
>

I am not a fan of removing context and nuance, and restating pieces in a
formalized way.

The problem with the Reply-To preference is that it is an emacs-vs-vi type
discussion.  It's been had for a long time, and there is no true right or
wrong.

What I did was illustrate how we historically arrived on this side of this
argument.  And how we ended up with this as a default and majority
configuration.  I mentioned modern email client behaviour.

As I indicated, we can obviously open up discussion on this topic - given
the above I wasn't sure it to be worth it.  FWIW, if there is a strong
feeling within this community that we really really need to change, I would
have expected that to have surfaced by now.  Clearly the door remains open
to still raise it back up.

Daniel
> (please Cc, as I'm not subscribed)


Cheers,

Sander

Re: Can we please disable harmful "Reply-to" munging ?

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Tristan Van Berkom wrote on Wed, Jul 08, 2020 at 23:00:36 +0900:
> On Tue, 2020-07-07 at 21:34 +0200, Sander Striker wrote:
> > Given the above, I'm inclined to just leave it as-is.  But obviously we
> > could discuss further if need be.
> 
> While I certainly don't agree with it, it would appear that the ASF has
> already taken a project-wide position on this,

The ASF does not have a cross-project position on Reply-To adding.

The ASF exists to enable projects to do their thing.  Among other
things, that means the ASF doesn't tell projects how to run their
mailing lists.  Project lists are created with Reply-To adding enabled
by default, but that's just that: a default, not a mandate.  It is
dev@foo.apache.org's decision whether or not dev@foo.apache.org should
set Reply-To.

(Personally, I would have expected the Apache lists to be created by
default with the same Reply-To settings as the respective pre-Apache
lists.  There used to be a checkbox for this in the mailing list
creation request form.)

---

Another point of view: ASF doesn't choose winners.  That's why ASF is
happy to host competing projects and is allergic to majority votes.  Any
sort of "Project dev@ lists MUST be configured to do X" policy would be
a choice of winner, and so would need very good reasons.

---

Not directly related, but there are at least three options for y'all to
choose from:

- Add no headers.
- Add Reply-To.
- Add Mail-Followup-To.

> > I remember from my early days in the ASF (2004) that we also had a short
> > conversation about Reply-To [4].  The expectation seemed to be that replies
> > went to the list.  And annoyance was voiced more regarding inconsistency
> > than anything else.  The result was the inconsistency being fixed.
> >

So, let's formalize your argument for a minute, Sander:

Premise: The Apache HTTP Server developers decided in 2004 to configure
their dev@ list in the same way their spun-off sister project's dev@
list was configured.

Conclusion: Buildstream should configure its lists the same way today.

I don't see how the conclusion follows from the premise.

Daniel
(please Cc, as I'm not subscribed)

Re: Can we please disable harmful "Reply-to" munging ?

Posted by Tristan Van Berkom <tr...@codethink.co.uk>.
Hi,

On Tue, 2020-07-07 at 21:34 +0200, Sander Striker wrote:
> Hi Tristan,
> 
> On Tue, Jul 7, 2020 at 11:26 AM Tristan Van Berkom <
> tristan.vanberkom@codethink.co.uk> wrote:
> 
> > Dear mailing list admins,
> > 
> > I've noticed a strange behavior while replying to emails on this new
> > list, as if the list is trying to control to whom I send the replies,
> > and upon inspection, I have verified that this list is indeed munging
> > the reply-to headers.
> > 
> > As this has been well known to be harmful for decades[0], I would ask,
> > can we please disable this reply-to munging on our mailing list ?
> > 
> 
> Obviously this has been a debate back when... The oldest instance I could
> find is from 1995(!) [1].
> 
> In the context of the ASF, where "If it didn't happen on-list, it didn't
> happen" (e.g. [2][3], the reply-to to the list can actually be seen as a
> feature, not a bug.  It makes it a conscious decision to not publicly reply.

Precisely this.

A conscious decision to post on-list cannot be damaging when you forget
to post to the list, while a conscious decision to reply privately can
be damaging when forgetting to explicitly reply privately.

> I've inquired with the Infrastructure team and it turns out that most
> (read: near all) of the lists have reply-to set to the list.  With the
> exception of commits@ lists, which typically have reply-to set to the dev@
> list.
> 
> I remember from my early days in the ASF (2004) that we also had a short
> conversation about Reply-To [4].  The expectation seemed to be that replies
> went to the list.  And annoyance was voiced more regarding inconsistency
> than anything else.  The result was the inconsistency being fixed.
> 
> Now, as you mention mail clients, it appears that the client I'm using at
> the moment (Gmail) will use the original sender when "Reply"-ing, and will
> use all To/Cc when "Reply All"-ing.  Also, a little birdie told me that
> Thunderbird also has this behavior with an additional "Reply List" function.
> 
> Given the above, I'm inclined to just leave it as-is.  But obviously we
> could discuss further if need be.

While I certainly don't agree with it, it would appear that the ASF has
already taken a project-wide position on this, which is probably not
worthwhile rehashing.

Cheers,
    -Tristan



Re: Can we please disable harmful "Reply-to" munging ?

Posted by Sander Striker <st...@apache.org>.
Hi Tristan,

On Tue, Jul 7, 2020 at 11:26 AM Tristan Van Berkom <
tristan.vanberkom@codethink.co.uk> wrote:

> Dear mailing list admins,
>
> I've noticed a strange behavior while replying to emails on this new
> list, as if the list is trying to control to whom I send the replies,
> and upon inspection, I have verified that this list is indeed munging
> the reply-to headers.
>
> As this has been well known to be harmful for decades[0], I would ask,
> can we please disable this reply-to munging on our mailing list ?
>

Obviously this has been a debate back when... The oldest instance I could
find is from 1995(!) [1].

In the context of the ASF, where "If it didn't happen on-list, it didn't
happen" (e.g. [2][3], the reply-to to the list can actually be seen as a
feature, not a bug.  It makes it a conscious decision to not publicly reply.

I've inquired with the Infrastructure team and it turns out that most
(read: near all) of the lists have reply-to set to the list.  With the
exception of commits@ lists, which typically have reply-to set to the dev@
list.

I remember from my early days in the ASF (2004) that we also had a short
conversation about Reply-To [4].  The expectation seemed to be that replies
went to the list.  And annoyance was voiced more regarding inconsistency
than anything else.  The result was the inconsistency being fixed.

Now, as you mention mail clients, it appears that the client I'm using at
the moment (Gmail) will use the original sender when "Reply"-ing, and will
use all To/Cc when "Reply All"-ing.  Also, a little birdie told me that
Thunderbird also has this behavior with an additional "Reply List" function.

Given the above, I'm inclined to just leave it as-is.  But obviously we
could discuss further if need be.

Cheers,

Sander

[1]
https://lists.apache.org/api/source.lua/431494706a6404b853e046f412d2f5a6351f9375300904c7d2415712@816231838@%3Cdev.httpd.apache.org%3E
[2] https://blogs.apache.org/foundation/entry/apache-in-2017-by-the
[3] https://blogs.apache.org/foundation/entry/celebrating_17_years_of_the
[4]
https://lists.apache.org/thread.html/1cc9bd09c220d3b62078e2f9fcefebd205fe79c0a3aee1e20ad25a95%401095938449%40%3Cdev.httpd.apache.org%3E

Especially for the sake of following the principles of least surprise
> and the least damage, as specified in the (very old) referred article,
> the inconvenience of occasionally receiving multiple copies of the same
> email (one to you, and one to the list, and only when using "reply all"
> instead of "reply to list" which some modern mail clients also now
> support), is a very small price to pay.
>
> Best regards,
>     -Tristan
>
> [0]: https://www.unicom.com/pw/reply-to-harmful.html
>
>
>