[news.eclipse.technology.ldt] Re: keeping an eye on the competition :-)

[Dave Wathen <dave.wathen@xxxxxxxxxxxx>]

Chris Daly wrote:
> One way to shape our thinking about what LDT could and should try to do 
> would be to look at what other IDEs are doing today in the 
> multi-language space and where they are going.
>
> What follows are some notes on projects I've been tracking that have to 
> do with IDEs that come with integrated language designers and language 
> tooling generators:
>
> - Sergey Dmitriev, of the IntelliJ IDE company JetBrains, recently wrote 
> a manifesto called "Language Oriented Programming" where he talks about 
> a new IDE he is developing that includes a "Structure Language" and an 
> "Editor Language".  These are languages designed to express languages 
> and their IDE tooling.
> http://www.onboard.jetbrains.com/is1/articles/04/10/lop/
> http://www.jetbrains.com/mps/
>
> - The company Xactium has a product called XMF-Mosaic which is sort of 
> an IDE for designing DSLs and building the IDE tooling for the DSLs you 
> design.  I really recommend their book "Applied Metamodeling" which you 
> can download for free from their site (go to Downloads area).
> http://albini.xactium.com/content/
>
> - Microsoft has a project called "DSL tools" for Visual Studio.  Look 
> for their developer blogs that talk about it and also look at the book 
> Software Factories (Greenfield et al.), especially chapter 8.
> http://lab.msdn.microsoft.com/teamsystem/community/blogs/
> http://blogs.msdn.com/stevecook
> http://blogs.msdn.com/stuart_kent
>
>
> These are just three examples and they surely reflect my bias towards 
> DSLs.  I'd be interested in hearing more from you all (I'm sure Tim and 
> the others from Borland could add something to this) about non-Eclipse 
> multi-language efforts that would be good models to study to help shape 
> the LDT.
>
>
> Chris

I thought these were very interesting links and I hope the LDT work can 
provide a foundation for this approach.  I think the idea of a language 
to model languages and tooling has to be the approach that provides the 
easiest implementation for later language projects (assuming it's done 
well).

I found the Applied Metamodelling book a useful but I thought the 
examples were overly simplistic (always hard to find the right balance 
thought) and I thought  it was too heavily slanted towards OO.  If 
you're going to support multiple languages you also need to address 
multiple paradigms.  I appreciate that the OO is being used to model 
other languages and could model (say) functional languages but it would 
have been nice to see some treatment of this.

Also I think the book misses the point when it talks about the process 
of language design.  If language development is to form a primary 
technique for software development it need to be more accessible than 
this to programmers who aren't prepared to immerse themselves in the 
theory.  The material on language extensions and reuse goes in the right 
direction for this.

I'd recommend taking a look at Mozart/Oz http://www.mozart-oz.org/ and 
there's a book to go with this 
too.http://www.amazon.co.uk/exec/obidos/ASIN/0262220695/qid=1115846440/sr=8-1/ref=sr_8_xs_ap_i1_xgl/202-8610107-3311826
(there a draft PDF version knocking around on the web somewhere too). 
This work shows a very highly abstracted treatment of programming 
languages from several paradigms.