Re: [cdt-dev] CDT Roadmap Proposal

["Alain Magloire" <alain@xxxxxxx>]

> 
> Also on the table is QNX's proposal to release 2.1 off of the main branch 
> by the end of October (the illustrious Halloween release!). You guys can 
> correct me if I miss anything but I believe this release will contain 
> improvements and completion of the type hierarchy and class browser 
> features introduced in 2.0 and some major debugger enhancements. As well 
> we may try to squeeze even more performance enhancements to the parser if 
> we have time after taking a breath from 2.0.1. There is also expressed 
> desire to bring some of the managed build enhancements to this release if 
> possible.
> 
> With those out of the way, I'd like to propose a somewhat regular schedule 
> that we can count on for our collective future product planning. Despite 
> objections, the feedback I've received from various sources is that it is 
> critical that we line up with major Eclipse platform releases. These 
> appear to be happening yearly. Eclipse 3.1 is scheduled for 2Q05 so I'd 
> like to propose that CDT 3.0 be scheduled to release on the same day as 
> Eclipse 3.1. This will require a good understanding of the changes Eclipse 
> is planning and a good working relationship with them to ensure we don't 
> repeat the Eclipse 3.0/CDT 2.0 experience. This release would have the 
> work by Intel and others on managed build as well as our planned scalable 
> AST (DOM) work. I'm sure we'll see other goodies like the remote systems 
> model.
> 

Agreed.

This is more a FYI, because we may not want to rush in Eclipse-3.1
until it reaches a certain stability.  But some folks may wonder why
there PR(s) xxyyzz were not fix for CDT-2.1:

This time around we are trying to work hard with the eclipse platform folks instead
of specific CDT workarounds.  The results are good... but ... some crucial improvements
are moved to Eclipse-3.1 to not destabilize the 3.0 maintenance stream.  This is reasonnable but
not so great for us, if we want to make a release CDT-2.1 in october.
That means that some "fixes" and enhancements will have to be left behind.

I'll give a scenario, the resolver Model:
One of the main problem with the ResolverModel is its non-iteraction with
the rest of eclipse. That means if you define *.tcc has a C++ source file,
doing CVS diff(StructurCreator), starting the editor(from the navigator view), team file types
etc ...  will not work as expected.  In Eclipse-3.0 the new scheme IContentManager was
incomplete and the fixes/improvements are schedule only for 3.1

Insert many other examples/scenarios(like code folding).

Anyway just a heads up and bringing down the expectations for 2.1