Open Stack

On 03/17/2014 08:07 PM, Philip Schwartz wrote:
>
>
> On 3/17/14, 5:33 PM, "Sean Dague" <sean at dague.net> wrote:
>
>>
>> Have you considered other open source efforts to build upon, like
>> phabricator? That came up on IRC a few nights ago by Ryan Lane. And it
>> seems like a lot of mileage could be gained from contributing to an
>> existing upstream, even if it's an alternative one to gerrit.

> I have nothing against starting from a known base if there is something
> that meets a majority of our needs and can be easily enhanced. With that
> said, I would not feel that Phabricator would fit into that. It is not bad
> at what it does, but happens to be a very odd PHP app. From my experience
> with PHP (which is vast), server side applications that do any
> functionality that is beyond being just a web app tend to have poor code
> bases and very strange hacks to get around the fastcgi or mod_php sandbox
> that PHP runs in.


I looked at Phabricator before storyboard - I agree with Philip. It's a 
good application, but I don't think it's a good application for us. Same 
as with bug tracking - we have some really specific requirements.

> I have been looking at a lot of things that all are code review related
> and also looked at review board as suggested earlier. All have elements
> that I like and do not like and none of them meet the needs of the
> OpenStack project completely. Personally I feel that making an attempt at
> meeting our needs while leveraging external libraries is good idea, and
> using things that applications like review board use and some of the
> libraries they use is a good starting point.

I think we have an opinionated view of how this works - and as it 
interfaces with things like zuul and turbo-hipster, we deal with a 
massive amount of things that are pretty specific to us. I'm not saying 
that's a thing to be proud of or a thing to be ashamed of - it simply is.

Which means, although I tend to have a side which agrees with Clint and 
Clark that replacing gerrit is a bit of a potential giant rathole - I 
also think that making a scalable thing that architecturally fits with 
the other things we've got would be nice. I don't really care about the 
java v. python part - but ultimately gerrit is designed as a single 
monolithic service - and although we haven't hit its ability to scale .. 
yet ... I think it's only a matter of time.

The google guys run their gerrit on a google specific sekrit backend 
after all.