On Wed, Mar 28, 2012 at 11:24, Vincent Massol
<vincent(a)massol.net> wrote:
Hi Denis,
On Mar 28, 2012, at 11:04 AM, Denis Gervalle wrote:
[snip]
> Denis, do you plan to refactor some existing
code to use this new API?
IMO
> this would be the best next step to prove
that it works.
It works, I use it everyday ! Their minor differences only impact really
special usage.
We all know the "it works on my machine" syndrome but that doesn't prove
anything… :)
As I said in another mail we really need a very code unit test coverage.
> Fo you have any idea about what code could be
migrated?
>
IMO, we may start using it for new components, or components that we are
refactoring, since it will allow being independent of the old core for
these component. This module may works for a while in parallel with the
existing service.
I was asking for an example of what we could start with. Do you have any
idea?
By merging it, and providing it as an option, it
will allows more user to
try it. There is a bridge between the old service and this new one that
you
may simply activate in xwiki.cfg. This new
service improve performance,
allow adding more rights on the fly and allow customizing the way it
works,
so it is an answer to those wanting these
improvements.
> We need to be careful not to keep this new security module not used for
> too long as otherwise it could "rot".
It will not stay unused on my side, so be sure I will stay close to it,
since I have written it because I need those improvements. I really
encourage anyone to try it !
> So IMO we need to start using it ASAP and then have a plan for switching
> to it.
>
IMO, the switch will be welcome when we have discussed the small
differences it has, and fix the few we do not tolerate. In particular,
the
startup of an empty wiki require login has
superadmin with this new
module.
I know this require more documentation to be
written, I am really aware
of
that, this just require time to be done... hopes
Andreas and others could
helps as well.
For me you are now responsible for this part (even though technically
we're all responsible of course since we use no code ownership ;) - but for
example Thomas and I know more the Rendering and are fixing stuff there,
while others have more knowledge in other parts).
Have I said the contrary ? All I said is that I will do my best supporting
it, but this is really a large part, and some help is also welcome. I have
taken the time to have it at the current state in the hope we will be able
to move out of the crappy service we use currently more that 20 times per
requests. It is now mature enough to start using it, since we do, and...