[Pnet-developers] Feature selection in pnetlib

[Rhys Weatherley]

Hi all,

Once upon a time, there were only two ways to build pnetlib.  ECMA_COMPAT and 
not-ECMA_COMPAT.  With the introduction of profiles, this has changed 
dramatically.  Individual features such as reflection, floating-point, 
serialization, etc, can be included or omitted.

Over the last few weeks, I've been going through the pnetlib code and marking 
up features with CONFIG_??? macros so as to make it easier to select them.  

This is an ongoing process, which all pnetlib developers need to be aware of.  
It isn't enough to just put "#if !ECMA_COMPAT" around non-ECMA things any 
more.  It is better to use a descriptive, feature-specific, label.  Even some 
ECMA things (like reflection) have labels now.  See "pnetlib/profiles/full" 
for a current list of feature labels.

I'm working my way through the namespaces, and if you want to help out then 
let me know and I'll assign CONFIG_??? labels for new features as required.  
We don't want to overdo it, of course.  Too many options will just create 
confusion.  Perhaps we can discuss this further at this weekend's meeting?

Where all this is heading is a flexible pnetlib that can be sliced and diced 
in multiple ways for different embedded system configurations without needing 
to fork the code to do it.  i.e. one code base with multiple personalities.

Cheers,

Rhys.