Re: [buckminster-dev] Dependency Visualization

[Dann Martens <dann@xxxxxxxxxxx>]

Hi Achim,

This discussion is not simple and potentially endless, and that's why I 
refuse to engage in it anymore.

I'm currently working on two other Eclipse-based projects and I need to 
focus. If the need arises to use EMF in those projects, I won't 
hesitate, I can assure you.

I was in need of a build system badly, and I'll continue my pursuit, 
although not necessarily along this track.

Thanks,
 Dann

Achim Demelt wrote:
> Dann,
>
> I'm not sure that what I'm going to write will change your mind, but I'll 
> try anyway.
>
> I am using both EMF and Buckminster very intensively. I have built products 
> based on EMF. Built, of course, with Buckminster. I can run EMF-based code 
> generators during my automated build with Buckminster. Yes, EMF is already 
> running in my headless Buckminster. So I guess I know what I'm talking 
> about.
>
> EMF will not introduce any significant overhead to a typical Buckminster 
> installation. The runtime JARs are _really_ small. And there's only few of 
> them. At runtime, I'd argue that it may even _decrease_ Buckminster's memory 
> footprint.
>
> Installation-wise, I predict that people will see less of these "cannot 
> install XY" problems with Eclipse 3.5. That's for two reasons: One, p2 has 
> matured a lot in 3.5. Two, people (i.e. plug-in providers) have learned how 
> to specify their dependencies correctly. They used to be sloppy with this, 
> and while the old update manager accepted a lot of crap, p2 threw their 
> failures into the faces of their users. I learned this, I'm sure others 
> have, too.
>
> I won't reiterate all the benefits that EMF will bring. But I hope that I 
> could mitigate some of your fears.
>
> Regards,
>  Achim
>
>
> Dann Martens wrote:
>
> > Here we go again :)
> >
> > I feel some of the reply you've formulated doesn't address what I've
> > tried to convey.
> >
> > As far as Buckminster is concerned, I have managed to build a complete
> > working build system with additional plug-ins without any need of EMF,
> > already two years ago. I see no compelling reason to introduce EMF now,
> > as what I feel is needed to further Buckminster is totally unrelated.
> > I'm interested in having Buckminster stable (which it isn't) and in
> > contributing back what I added myself at that point. I really see no
> > reason at all to warrant this diversion. This has nothing to do with the
> > qualities of EMF itself, but I will not support this decision.
> >
> > We can argue and misunderstand each other as much as you want, but this
> > feels like a real waste of my time. And I'm sure yours as well. For a
> > silly build tool, no less.
> >
> > Best regards,
> >   Dann