Thank you for this answers, I will try

2007/3/8, Andreas Raab <andreas.raab at gmx.de>:
> Interesting. This is one of the cases where I fundamentally disagree
> with Bert's solution. Not because I don't have the "move problem" (I
> sure as hell do) but rather because of the troubles that are introduced
> by the out-of-order loading that happens with those modifications. For
> example, Tweak has a package "Scripting-Preload" that is required to be
> loaded *before* the "Scripting" packages. Works great if you load the
> packages independently but it can completely (and non-deterministically
> as we found out!) screw up if you load them via configurations. The
> whole idea of having a different load-order when you load the
> configuration vs. when loading the individual packages is extremely
> troublesome to me. It means that (due to the non-deterministic
> load-order) you cannot possibly predict the order in which things get
> loaded/initialized, which also means that given wide enough
> distribution, there is really *no* workaround for the initialization
> problem (since you cannot test in which order things get loaded) which
> is what ultimately led me to revert the changes in Croquet's MC variant.
>
> Cheers,
>    - Andreas
>
> Ron Teitelbaum wrote:
> > I definitely agree with Bert.  This really is a terrific solution and I have
> > found it to be very useful.  When I find a problem in code, I fix it and
> > store the next numbered version in my own repository.  My code loads
> > automatically until a new version of the package is released, hopefully with
> > my suggestions added.  It's really cool.
> >
> > Ron
> >
> >> -----Original Message-----
> >> From: squeak-dev-bounces at lists.squeakfoundation.org [mailto:squeak-dev-
> >> bounces at lists.squeakfoundation.org] On Behalf Of Bert Freudenberg
> >> Sent: Wednesday, March 07, 2007 6:21 PM
> >> To: The general-purpose Squeak developers list
> >> Subject: Re: How-to Monticello Configuration
> >>
> >>
> >> On Mar 8, 2007, at 0:07 , Philippe Marschall wrote:
> >>
> >>> 2007/3/7, Bert Freudenberg <bert at freudenbergs.de>:
> >>>> On Mar 7, 2007, at 23:37 , Philippe Marschall wrote:
> >>>>
> >>>>> With MCC the specified Monticello versions are loaded in the given
> >>>>> order from the given repositories.
> >>>> Well, actually, I changed that about a year ago in the impara
> >>>> version:
> >>>>
> >>>>         http://source.impara.de/mc.html
> >>>>
> >>>> It solves the problem of moving stuff between packages. Without that,
> >>>> moving methods or classes between packages is impossible to do
> >>>> correctly.
> >>> But creates a new one when a package assumes an other package is
> >>> present and initialized. As a result strange things may happen in
> >>> class side #initializes, and actually did happen for some of our
> >>> projects so we reverted that behavior.
> >> Well, there are workarounds for the init problem. There is no
> >> workaround for the moving problem, except for manually fixing up
> >> things when it's to late. So I'd opt for the former - but YMMV of
> >> course.
> >>
> >> - Bert -
> >>
> >>
> >>
> >
> >
> >
> >
>
>
>


-- 
Damien Cassou