Killing the gerrit to dev-list spam ...

[Lionel Elie Mamane]

On Wed, May 01, 2013 at 09:33:49PM +0200, Bjoern Michaelsen wrote:

> Looking at:

>  http://nabble.documentfoundation.org/Dev-f1639786.html

> depending on time of day you find 50-90% patch mails on there. As the writer of
> mail forward I have been repeatedly asked if we can kill this spam.

> I fully agree with this sentiment (...)

> So, is there anyone who is _not_ a sponsored developer opposing to
> kill the mail spammage?

I'm unhappy about the current situation, but in the *other*
direction. The most important part of gerrit is the *discussion* on
patches, and that's absent from the current mails; having to
constantly go to a browser is disruptive; Mail / mailing lists have
all kind of tools that I can use to manage what I'm interested in and
what not, from e.g. "ignore this thread", "watch this thread", etc,
which gerrit only imperfectly has. Even more basically: when it is a
mailing list, my MUA keeps track for me of what I have read and what I
have not read yet, which gerrit does not do at all.

(Never mind that the tools I have to manage mailing list mails are
 under my control and choice: I can patch my MUA locally, but I'm
 dependent on what the centrally installed gerrit offers as services.)

So, that's "IMO"; obviously a ML is a shared resource, so if the
others disagree...

> - could this be mitigated by a separate gerrit-patches mailing list?

Yes. I would call it "patch-discuss" or "code-discuss" rather than
"gerrit-patches"; gerrit is just *one* technical way to submit /
discuss patches, there is AFAIK no reason to separate "gerrit" patches
from "sent to Mailing List" patches.

Actually, separating:

1) discussion on "exact code" (a concrete patch) on the one hand, and

2) general design discussions, process discussions, "hi I'm a new
   developer" mail, etc.

into two different mailing lists could make sense. That is
generalising a bit your proposal, but to me it makes sense :)

> - could this be mitigated by a daily digest of "gerrit news"?

If I can "undigest" it automatically on my side, I don't mind.

> - could this be mitigated by other means?

Probably.

> Not that this does:

> - not mean that it is evil to send a patch to the list (although its a bit
>   misguided given how gerrit simplifies and enables things ;) )

It would make more sense to me that a patch manually sent to "the
list" goes to the same list as the gerrit patches.

-- 
Lionel