[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [gmt-dev] Initial UMLX prototype
|
Hi,
I've added my thoughts below
Jeff
On Tue, 03 Jun 2003 22:23, Willink, Ed wrote:
> Hi All
>
> Apologies for any mis-understanding, but as far as I can see I did what
> was agreed. All the UMLX compiler and language stuff is under UMLX.
> Non-UMLX stuff is in re-usable directories. The sole exception is
> UmlxDecorator that sits more comfortably as an XmiEmitter sibling with
> GME-specific stuff.
>
> If you insist that every sub-project, put all its external support stuff
> under
> xyz we will end up with a zillion copies of libraries and tools.
Perhaps tools that are common should go into some "common" area/project. This
would be only if they are common to most sub-projects and not just common to
a single sub-project. Perhaps its even conceivable that some sub-projects
within GMT may depend on different versions of some tools (hopefully not), or
libraries specific to the sub-project (which implies a lib directory in the
sub-project). In large projects its not uncommon to have mini-projects that
have a similar structure to the overall project, which are built and pulled
together. This being possible perhaps if UMLX started as a sub-project then
tools/components that were required by other GMT projects could be "promoted"
to the higher level?
>
> For instance, should specialized Sun Java libraries be in sub-projects?
Ah no this would be silly wouldn't it? I assume that we would just list this
as a prerequisite that ppl should obtain (link provided), just as we would
list any other such requirements.
>
> NiceXSL is a preprocessor for XSLT. It is nothing to do with UMLX.
Again, perhaps we should list this as a prerequisite with a link to the
download location?
>
> XmiEmitter is an add-in for GME, it is nothing to do with UMLX.
>
> GME is a third party tool - nothing to do with UMLX.
People should download GME separately then.
>
> If you really want UmlxDecorator moved fine, but moving the rest is a
> nonsense.
>
> In general, moving anything within a Code Management System is very
> undesirable,
> so I'm very reluctant to move anything now.
Hence the idea to get agreement before checking anything when the components
are not yet agreed. I think I also provided a link to some project
guidelines, amongst them was information on how write access to CVS was
controlled this may also affect the structure, if you want to have controlled
access to specific areas within GMT. I assume that anyone of the core
contributors (if they all belong to the same group) can delete the stuff
(current structure) from the repository, so that you can check it back in.
>
> Ed
>
> > -----Original Message-----
> > From: Jorn Bettin [mailto:jorn.bettin@xxxxxxxxxxxxxxxx]
> > Sent: Tuesday, June 03, 2003 8:18 AM
> > To: gmt-dev@xxxxxxxxxxx
> > Subject: Re: [gmt-dev] Initial UMLX prototype
> >
> >
> > Hi Ed,
> >
> > Please relocate all the sub-directories <sd> - except /umlx -
> > that you've
> > created under org.eclipse.gmt to org.eclipse.gmt/umlx/<sd>
> >
> > Thanks,
> > Jorn
> >
> > ----- Original Message -----
> > From: "Jeff Hoare" <jhoare@xxxxxxxxxxxxxxxx>
> > To: <gmt-dev@xxxxxxxxxxx>; "Willink, Ed" <Ed.Willink@xxxxxxxxxxxxxxx>
> > Sent: Tuesday, June 03, 2003 8:26 AM
> > Subject: Re: [gmt-dev] Initial UMLX prototype
> >
> >
> > Hi,
> > I thought that all the umlx code was to be placed (for now)
> > under a umlx
> > sub-directory and to be shifted to appropriate locations at a
> > later date
> > (under an agreed component structure). However, you appear to
> > have created a
> > directory structure along the lines you proposed in an
> > earlier email. Am I
> > missing something where you all agreed to Ed's structure.
> >
> > Jeff
> >
> > On Mon, 02 Jun 2003 20:37, Willink, Ed wrote:
> > > Hi All
> > >
> > > I have put all my UMLX code, support tools and some models into CVS.
> > >
> > > org.eclipse.gmt/umlx/ReadMe gives a few clues on installation.
> > >
> > > org.eclipse.gmt/umlx/examples/AddressBook/ReadMe gives a little
> > > overview on the status to date of the example in the
> > > org.eclipse.gmt-home/doc/umlx/UmlxGrpahicalLanguage.pdf
> > > paper.
> > >
> > > So that the example can be studied without any installation,
> > > a load of intermediate and output files are also provided.
> > >
> > > NB. The example is still in progress, but it works except for
> > > ignoring the rather important constraint between the mdels,
> > > as a result of which the updated address book is just a copy.
> > > Hope to get this fixed this week.
> > >
> > > Probably the Transformation.xmi is most ineresting because it
> > > shows the intermediate scan and build models.
> > >
> > > Design 'documentation' in umlx/models/UmlxCompiler.mga is
> > > annoitated to make it more readable in umlx/models/UmlxCompiler.doc.
> > > The implementation in umlx/bootstrap has close but far from precise
> > > correlation with the documentation.
> > >
> > > Getting the example in to place with two inputs forced me to clean
> > > up a number of outstanding niggles. I had previously always used
> > > a single GME file for conveniemce. Splitting schema, transforms
> > > and usage into separate sessions caused a problem with GME, the
> > > schema nodes when referenced from transforms and from usage had
> > > distinct xmi:id's, so I had to reconstruct them all from
> > > hierarchical names, so what was a slightly annoying 20 character id
> > > from GME is now a really annoying 100 character path, but at least
> > > it's full of diagnostic value. Compressing this is yet another
> > > optimisation pass.
> > >
> > > Note that the XMI is XMI 2.0, which very few tools support yet. we
> > > may want to write a trivial XMI 1.x to 2.0 transform.
> > >
> > > [A tip from the Ptolemy group. Make sure you have set your
> > > Preferences->CVS to maximum 9 compresoion, unless you've got a
> > > really good dial-up speed.]
> > >
> > > Regards
> > >
> > > Ed Willink
> >
> > --------------------------------------------------------------
> > ----------
> >
> > > E.D.Willink, Email:
>
> mailto:EdWillink@xxxxxxx
>
> > Thales Research and Technology (UK) Ltd, Tel: +44 118 923 8278 (direct)
> > Worton Drive, or +44 118 986 8601 (ext 8278)
> > Worton Grange Business Park, Fax: +44 118 923 8399
> > Reading, RG2 0SB
> > ENGLAND http://www.computing.surrey.ac.uk/personal/pg/E.Willink
> > ------------------------------------------------------------------------
> > (formerly Racal Research and Thomson-CSF)
> >
> > As the originator, I grant the addressee permission to divulge
> > this email and any files transmitted with it to relevant third parties.
> > _______________________________________________
> > gmt-dev mailing list
> > gmt-dev@xxxxxxxxxxx
> > http://dev.eclipse.org/mailman/listinfo/gmt-dev
>
> _______________________________________________
> gmt-dev mailing list
> gmt-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/gmt-dev
>
>
> _______________________________________________
> gmt-dev mailing list
> gmt-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/gmt-dev
> _______________________________________________
> gmt-dev mailing list
> gmt-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/gmt-dev
