Multiple identical mails from Gerrit

Multiple identical mails from Gerrit

[Eli Zaretskii]

Is there any reason why some (or all?) email messages from Gerrit
arrive at my inbox several times?  For example, this message:

  From: "Sourceware to Gerrit sync (Code Review)" <gerrit@gnutoolchain-gerrit.osci.io>
  To: Christian Biesinger <cbiesinger@google.com>,	Simon Marchi <simon.marchi@polymtl.ca>, gdb-patches@sourceware.org
  Auto-Submitted: auto-generated
  Subject: [pushed] Replace SYMBOL_*_NAME accessors with member functions

arrived as 3 identical messages.

Why does this happen?

Re: Multiple identical mails from Gerrit

[Simon Marchi]

On 2019-11-22 2:19 p.m., Eli Zaretskii wrote:
> Is there any reason why some (or all?) email messages from Gerrit
> arrive at my inbox several times?  For example, this message:
> 
>   From: "Sourceware to Gerrit sync (Code Review)" <gerrit@gnutoolchain-gerrit.osci.io>
>   To: Christian Biesinger <cbiesinger@google.com>,	Simon Marchi <simon.marchi@polymtl.ca>, gdb-patches@sourceware.org
>   Auto-Submitted: auto-generated
>   Subject: [pushed] Replace SYMBOL_*_NAME accessors with member functions
> 
> arrived as 3 identical messages.
> 
> Why does this happen?
> 

When a patch with a Change-Id corresponding to an open Gerrit change is manually
pushed to the binutils-gdb git repo, Gerrit notices it and that automatically
creates a new version of the change, reflecting what was actually pushed.  That
sends a mail of type "new patchset version".  It then changes the state to
"Closed", which sends another email.

When I'm not subscribed to a particular change (because I'm not a reviewer or in
the CC list), I receive these two emails from the list.  When I am subscribed to
a change, I receive these two emails twice (once through the list and once
directly to my email), for a total of 4.

I don't know why you have 3 messages for this change, this is an odd number, and
you are not subscribed to it either...

Simon

Re: Multiple identical mails from Gerrit

[Eli Zaretskii]

> From: Simon Marchi <simark@simark.ca>
> Date: Fri, 22 Nov 2019 15:53:42 -0500
> 
> When a patch with a Change-Id corresponding to an open Gerrit change is manually
> pushed to the binutils-gdb git repo, Gerrit notices it and that automatically
> creates a new version of the change, reflecting what was actually pushed.  That
> sends a mail of type "new patchset version".  It then changes the state to
> "Closed", which sends another email.

Why do these trigger sending email at all?  We didn't get email about
pushing changesets before Gerrit, so why do we need them now?

> When I'm not subscribed to a particular change (because I'm not a reviewer or in
> the CC list), I receive these two emails from the list.  When I am subscribed to
> a change, I receive these two emails twice (once through the list and once
> directly to my email), for a total of 4.
> 
> I don't know why you have 3 messages for this change, this is an odd number, and
> you are not subscribed to it either...

Maybe I made a mistake counting emails, there were several similar
ones.

Re: Multiple identical mails from Gerrit

[Joseph Myers]

On Fri, 22 Nov 2019, Eli Zaretskii wrote:

> > From: Simon Marchi <simark@simark.ca>
> > Date: Fri, 22 Nov 2019 15:53:42 -0500
> > 
> > When a patch with a Change-Id corresponding to an open Gerrit change is manually
> > pushed to the binutils-gdb git repo, Gerrit notices it and that automatically
> > creates a new version of the change, reflecting what was actually pushed.  That
> > sends a mail of type "new patchset version".  It then changes the state to
> > "Closed", which sends another email.
> 
> Why do these trigger sending email at all?  We didn't get email about
> pushing changesets before Gerrit, so why do we need them now?

On the glibc side, I tend to agree with that - people wanting to see what 
was pushed to master can follow glibc-cvs.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: Multiple identical mails from Gerrit

[Simon Marchi]

On 2019-11-22 4:11 p.m., Joseph Myers wrote:
> On Fri, 22 Nov 2019, Eli Zaretskii wrote:
> 
>>> From: Simon Marchi <simark@simark.ca>
>>> Date: Fri, 22 Nov 2019 15:53:42 -0500
>>>
>>> When a patch with a Change-Id corresponding to an open Gerrit change is manually
>>> pushed to the binutils-gdb git repo, Gerrit notices it and that automatically
>>> creates a new version of the change, reflecting what was actually pushed.  That
>>> sends a mail of type "new patchset version".  It then changes the state to
>>> "Closed", which sends another email.
>>
>> Why do these trigger sending email at all?  We didn't get email about
>> pushing changesets before Gerrit, so why do we need them now?
>
> On the glibc side, I tend to agree with that - people wanting to see what 
> was pushed to master can follow glibc-cvs.

I think that having a notification saying that the patch was pushed is nice.  People
reply "I pushed the patch",  so we might as well let the system do it automatically
anyway.

I could maybe get rid of the second one (state change), I'll have to look in the configuration.

Anyway: https://sourceware.org/ml/gdb-patches/2019-11/msg00749.html

Simon

Re: Multiple identical mails from Gerrit

[Eli Zaretskii]

> Cc: gdb-patches@sourceware.org
> From: Simon Marchi <simark@simark.ca>
> Date: Fri, 22 Nov 2019 17:03:54 -0500
> 
> >> Why do these trigger sending email at all?  We didn't get email about
> >> pushing changesets before Gerrit, so why do we need them now?
> >
> > On the glibc side, I tend to agree with that - people wanting to see what 
> > was pushed to master can follow glibc-cvs.
> 
> I think that having a notification saying that the patch was pushed is nice.  People
> reply "I pushed the patch",  so we might as well let the system do it automatically
> anyway.

My problem with these messages is that I have to look at the fine
details of the Subject line, and perhaps also at the beginning of the
body, to understand whether I need to review the patch.  Since I only
review the documentation parts, and rarely look at other parts of the
changeset, it is important for me to be able to quickly understand
whether a posted patch needs my attention.  And these messages, which
look almost exactly like an RFA but for minor details, make this
harder.

But maybe I'm the only one who is in this situation, in which case
I'll have to live with this.

> I could maybe get rid of the second one (state change), I'll have to look in the configuration.

Thanks you.

> Anyway: https://sourceware.org/ml/gdb-patches/2019-11/msg00749.html

Yep.  So maybe this issue is of academic importance ;-)