Re: FYI: Status of gdb usage of gerrit

[Joseph Myers <joseph at codesourcery dot com>]

On Wed, 27 Nov 2019, Carlos O'Donell wrote:

> I was hoping gerrit's email integration for feedback from the ML
> would work better than it does. Right now that's a showstopper for
> immediate use in that we want *one* system of record that works well
> with email responses.

There are a couple of key issues there (handling of replies to the initial 
notification of a change, handling of replies not quoting the footer 
information by using References headers instead).  I didn't see any 
evidence that those issues would be unfixable or that the gerrit 
maintainers objected to the idea of improving things in that area.

I'm not expecting any review system to be good out of the box, without 
some custom configuration and quite likely adding a few new features to do 
things we want (such as how the gerrit email notifications were improved 
to include diffs and more context in general).  We've identified some 
issues with gerrit that are things to consider when comparing it with 
other systems - not issues that mean it has to be discarded from 
consideration, if they can be fixed in an upstream-suitable way.  (I don't 
know how hard the issues Florian listed with the web interface might be to 
address, however.)

I'm pretty sure that when we started using Bugzilla for GCC and 
sourceware, various changes to the code were needed to make it work well 
for us.

-- 
Joseph S. Myers
joseph@codesourcery.com