Re: FYI: Status of gdb usage of gerrit

[Florian Weimer <fw at deneb dot enyo dot de>]

* Carlos O'Donell:

> On 11/26/19 5:24 PM, Joseph Myers wrote:
>> On Tue, 26 Nov 2019, Florian Weimer wrote:
>> 
>>> * Joseph Myers:
>>>
>>>> I don't think we should require pushing through Gerrit.  For changes not 
>>>> needing review it's an unnecessary complication; a normal git push ought 
>>>> to be fine and Gerrit ought to be able to handle it.
>>>
>>> Jonathan Nieder's comments
>>>
>>>   <https://sourceware.org/ml/libc-alpha/2019-11/msg00532.html>
>>>   <https://sourceware.org/ml/libc-alpha/2019-11/msg00533.html>
>>>
>>> suggest to me that whis would need custom Gerrit development (listed
>>> as option 3, “something fancy with two-way sync”).
>> 
>> No, just running Gerrit on sourceware - the first of those messages 
>> explicitly says "It copes fine with the repository being updated behind 
>> its back".  Running it on sourceware seems obviously the right thing to do 
>> for non-experimental use.
>  
> Correct.
>
> For changes pushed "behind" gerrit you would have to add the "Reviewed-by"
> lines and it would cause the gerrit review to be stuck in an intermediate
> state requiring cleanup because the commit message changed without review.
>
> For changes pushed "with" gerrit it would work just fine.
>
> So my suggestion was:
> * You review via gerrit and push in gerrit
> or
> * You review via email and push directly (no Changeset-Id).

I see.

I think this dual setup is too confusing.  People will ack patches on
one side of the fence that have been vetoed on the other side.

Are we still seriously considering using Gerrit? Personally, my plan
was to drain the patches currently in Gerrit and stop using it.

Should we try something else instead?

I wonder if I should put some effort into teaching Gnus how to post
patch series and import posted patch series into local branches.
Someone else should do the same thing for mutt.  But I don't know how
we can reach Thunderbird and webmail users.

The nice incremental diff review feature we've seen in Gerrit seems to
have recently landed in Git itself, as git range-diff.  I haven't used
it yet, admittedly.  However, I'm pretty much certain that better
local mail client integration would give us many of the benefits that
we've seen in Gerrit.