On Mon, 2010-09-06 at 17:42 -0500, Chris Muller wrote:
> Hi Ross,  your first instinct was right.  You need to load the
> pre-requisite packages first, but you just did not go far enough down
> the prerequisite tree.  Magma client has several of its own pre-reqs..
> 
> Quick question:  Does Cuis support installation of SAR files?  If so,
> you might try installing the latest SAR file downloadable from
> SqueakMap.
I pointed it at a downloaded sar file and it didn't seem to know what
too do.
> 
> Even if not, the SAR file has all of the packages and a plain-text
> loadable script inside to load them all in sequence.
I guess the MC packages have the information, or perhaps just the
definitions, in the big bin file they contain.
> 
> The full package list is also in the 1.1r2 announcement:
> 
>   http://lists.squeakfoundation.org/pipermail/magma/2010-July/001578.html
Thanks.
Ross
> 
> HTH,
>   Chris
> 
> 
> On Sun, Sep 5, 2010 at 5:13 PM, Ross Boylan
> <RossBoylan at stanfordalumni.org> wrote:
> > I got Magma tester-cmm.368.mcz from squeak source, extracted the
> > source.st file inside of it, and attempted to file source.st into a Cuis
> > 2.7 image.
> >
> > This fails early on because the file has
> > !MagmaRepositoryController methodsFor: '*magma tester' stamp: 'cmm 9/6/2007 21:44'!
> > very early on.  MagmaRepositoryController is undefined.
> >
> > I presume that either the Monticello machinery, or the way mainline
> > squeak handles this, makes this OK in the usual way of doing things.
> >
> > Guessing that the issue might simply be one of dependencies, I tried the
> > same steps with Magma client-cmm.490.st.  That failed similary with
> > MaVariableWordBuffer.
> >
> > Possibly resequencing the input so that the class definitions occur
> > first would work around this problem.  I'm a little surprised the
> > changesets aren't in that form already.
> >
> > Mostly these tests were in an image that included OSProcess.  I did one
> > test in a clean image; it failed the same way.
> >
> > My hope was that the file in's would work, that OSProcess would be good
> > enough to run the magma tests, and that the magma tests would pass.
> >
> > Ross
> >
> >
> >
>