[Mageia-sysadm] [186] Add review board to isntall

[Michael scherer]

On Sun, Nov 07, 2010 at 09:40:19AM +0100, Dexter Morgan wrote:
> On Sun, Nov 7, 2010 at 3:23 AM, Michael scherer <misc at zarb.org> wrote:
> > On Sat, Nov 06, 2010 at 09:31:35PM +0100, root at mageia.org wrote:
> >> Revision: 186
> >> Author:   dmorgan
> >> Date:     2010-11-06 21:31:35 +0100 (Sat, 06 Nov 2010)
> >> Log Message:
> >> -----------
> >> Add review board to isntall
> >
> > who asked for it ?
> > i do not think this was discussed, and given the current type of
> > contribution we have, I am not sure it will be suiable. At least,
> > I think this should first be discussed with packager, no ( and also
> > with sysadmin if we find it suitable, maintainable, etc )
> 
> i addedd it to not forgot, but reviewboard is something we _must_ have
> to allow more contribution to our own projects, tools, code, ...
> This is much better to use this than sending a patch on bugzilla.

A tool do not replace a process. If there is no one to do code review,
the tool will not be used. We cannot dictate "now, we do code review" without 
discussing with others involved persons first ( ie, coders, maybe mackagers ).
Maybe they would prefer a different way of doing it ( like email, like doing it on irc,
like doing it like postgresql ). If we want to empower the community as a whole, 
the community must do its own choices.

I do think code review is a good idea. And I do like reviewboard, don't get me wrong.
But before installing it and using it, others peoples must see how it can be useful, 
how it work, etc.

This also mean to agree on procedure, ie, do we let people propose patch on bugzilla ?
Who can review ? Who has the final word to say ? 

Ie, the tool must be derivated by the workflow, not the reverse. So first a workflow 
must be found, IMHO. 

And all of this requires to have the team in place, and to have someone involved
into trying to convince people of using best practices. Even if we will 
likely be right in the end, what is needed is that procedure like this are feel
welcomed by contributers, rather than imposed. This is indeed a more lengthy process, yes
and this is frustrating, yes. And this also may end differently that what we would
want, yes. 

> 
> what do you mean suitable ?
> Projects like kde use it since a long time so they have already tested
> and we would know if there were any maintainance issue

Well, transifex is also used since a long time by other projects ( Fedora for
example ), and no one raised the maintenace issues here afaik, 
despites them being know ( at least in the pre 1.0 version ). 

We also need to check if it support ldap and others things : 
- users in ldap, 
- dynamic acls based on ldap, ( ie, let all people in the sysadm team review patch for sysadm )
- a non webbased config (for puppet integration, or we would lose auditing and the advantage of vcs ).

1 seems ok. 2 and 3, rather not :/ 

( and this make me think this may not be the case for transifex etheir, and that we 
should have checked ). 
-- 
Michael scherer