Re: Google Gmail mailing list bounces

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Bob Proulx <bob@proulx.com> writes:

> Nate Bass wrote:
>> I compared my Gmail to the archives and I found that I wasn't
>> receiving mail from one particular address moasenwood@zoho.eu and I
>> think this caused me to bounce off the help-gnu-emacs mailing list.
>
> That is quite interesting.  Mail from most members came through okay.
> But mail from moasenwood@zoho.eu was frequently rejected by Gmail.  I
> don't see anything unusual about that zoho.eu (no DKIM for example)
> that would cause any difficulty.
>
>> This issue raises two questions.
>> 1) Why was I not receiving message from that address. Any thoughts
>> Emanuel? I did receive your original post "Gmane up again" and some
>> but not all of the following replies, if that helps.
>> 2) Should the automatic bounce processing system be changed? I think
>> my experience confirms that help-gnu-emacs has a bounce score
>> threshold of 5.
>
> Mailman has been a difficult program to improve.  The upstream has
> forked Mailman 2 into a rewrite that is quite different Mailman 3.  No
> one that I know has been running Mailman 3 yet.  So there is no
> experience with it.  However being a complete rewrite from the ground
> up I am confident predicting that it will have undesired behavior just
> like Mailman 2 and what that will be can't be predicted until it is in
> production service.  We will find out at that time.  Therefore Mailman
> 2 isn't getting design changes anymore.  Plus the instance on
> lists.gnu.org only gets updates "infrequently".  That's probably the
> best way to describe it.  It really depends upon who is employed as
> the admins at the time and that changes from year to year.

I upgraded a (very small-scale) Mailman 2 installation to Mailman 3, and
regretted it. I think Mailman 3 will be fine when it's ready, but my
impression is that it's not ready. Specifically, while it seems like the
core of the system (actually sending messages) works fine, the web
interface -- and the interaction between core at the web interface -- is
not there. Just lots of odd errors and mismatches, and rough patches.

I didn't use it long enough to really know if core is stable, mind
you -- it's just that in the time before I gave up on it, I didn't see
any problems related to core.

Also, much as I love Python, I wish they'd written it in something else.
Django for the web interface makes sense, but for core, it seems like
anything compiled would have been better. It's a huge memory hog.

Wish I could like it more...
Eric