[dsdp-mtj-dev] Initial thoughts on the fragmentation document

[Craig Setera <craigjunk@xxxxxxxxxx>]

I've spent some time looking at the document in the wiki on fragmentation
(http://wiki.eclipse.org/images/2/2e/MTJ_JME_Fragmentation_taks.pdf).  I 
think
it is a great start.  My primary concerns with the document in the have 
to do with
the categorization of the functionality.  I believe we need to have more 
categories
and that certain (current) products will actually fit into multiple 
categories.

My initial thought is that we need to have (at least) the following 
categories:

Fragmentation Source Language
------------------------------
This category would define the type and substance of the source code 
language
used within the Java code to differentiate "blocks" of code to be compiled
or not compiled.  Some entries in this category:

- Special comment blocks
   - Antenna Language
       - EclipseME
       - Antenna
       - AntME
   - JME Polish Language
- Annotation based language
- Others?

Processing Time
------------------
This is an implementation category.  Preprocessing operates on the Java
source while post-processing operates on the compiled classes and other
resources.

- Pre-processing
   - EclipseME
   - Antenna
   - AntME
- Post-processing
   - TIRA Wireless?

Embedded/External Processing
-----------------------------
Whether the processing is embedded as an Eclipse builder or is an external
builder such as Ant.

Embedded Builders:
- EclipseME

External Builders:
- Antenna
- AntME

Fragmentation Metadata
-----------------------
I agree that this is a valid and useful category.

====================================================================

I'm not really sure what to think of the "visual" category.  In the end,
this resolves down to the underlying editor support.  This is going to be
dependent on the fragmentation source language.  Some of this may be
reusable between the different languages, but I'm afraid there may be
very little reusable.  Is there already support from JDT for adding editor
support for fragmentation languages or will those features need to be
added?

The Ant-based services are currently defined to include EclipseME which is
not really the case.  In reality, this category as described feels to me
like a combination of the processing time and embedded/external processing
categories that I call out above.

I believe the post-processing services may be used as described in the 
document.
I think it is also valid to say that post-processing could be used instead
of pre-processing to alter the classes to be deployed.

At this point, these are my primary thoughts as related to the document.
I really appreciate your work in getting this initial document together and
I look forward to continuing this conversation with everyone.

Thanks,
Craig
EclipseME Lead Developer