[cdt-dev] Next steps for the CDT

[Sebastien Marineau <sebastien@xxxxxxx>]

Hi all,

jus thought I'd give a quick update on the CDT, where we are
at and what the next steps are...


1.0 release
-----------

CDT version 1.0 was released on November 11th. It is available for
Windows, Linux, QNX and Solaris hosts. Congratulations and thanks to
the whole CDT team, as well as all other folks who have helped out!
CDT 1.0 is an important milestone, and has generated a lot of interest
from other companies and enthusiasts.


1.0.1 Maintenance release
-------------------------

Based on the testing that occurred pre- and post 1.0 release, a number
of issues have been addressed, and some missing functionality has been
added. As we discussed before the 1.0 release, the proposal was to have
a "maintenance" 1.0.1 release shortly (a month or two) after the 1.0 GA.
This proposal is still in place; we are targeting mid-December as the date
to have a 1.0.1 candidate finalized. We'll be sending out a separate
email with the current proposed scope for 1.0.1 (bug fixes, extra features).
The good news is that the number of users of CDT has gone up, quite
a bit of testing (on small and large projects) has been going on, and
most of the reported issues to this point have been addressed. 


Interest from other companies
-----------------------------

On top of QNX, Rational, Redhat and Montavista, other companies have
shown interest or are working on products based on the CDT. These 
include Interstron, Timesys and Tensilica, which have been actively
building products based on the CDT, as well as others that have expressed
interest in getting more involved (SlickEdit, Wasabi, Macraigor, 
Programming Research to name a few). This is a very positive 
development and bodes well for the future of the CDT. It will, however,
place additional demands on the core CDT team to ensure
interoperability (through API's and extension points) of 
different plugins and products based on the CDT.


Planning
--------

Our original plans (from last August) had a CDT 1.0 release scheduled for 
end of October, a second major release scheduled for March and a third
scheduled for August. As we progress, gain experience and receive
feedback from users, the original plans will no doubt have to be 
updated. This has already happened with the provision for a 1.0.1 
release. 

Based on feedback from the participating companies, newsgroups and users, 
I see four major areas of focus in the next 6 months:

- Focus on the core plugins to implement API's and functionality.
Particular areas of focus include parsing and project indexing,
the build model and C++ support (class browser). We also need to
finalize the API's and extension points for the core.

- The debugger has generated a lot of interest, and has been the 
component whose configuration is most "tweaked". We'll need to focus
on having finalized API's for the CDI, as well as clearly documenting
how to extend the launching and debugging to suit variants of gdb.

- Documentation and testing and two areas that are still weak. Most
important is user documentation, which becomes important as we gain
more users. As for testing, we have in place some automated testing 
with JUnit, but this is still lacking in coverage (admittedly, there's
a lot of existing features and this is not an easy task). QNX and
RedHat have also been doing manual testing of the features and UI
components; it would probably be worthwhile to explore the possibility
of using some UI automation test tools.

- Improve the usability/usefulness of the CDT in a GNU/Linux environment
by developing the plugins to implement the GNUDE (e.g. configure/makefile 
generation, code assist, integration with help). This would not only
improve the experience for GNU/Linux users, but would also provide examples
of how to extend the CDT.


Next steps
----------

As the group of people involved in the CDT grows larger, it becomes
more difficult to do planning with in-person meetings. My suggestion
is that we kick off a round of planning for the next major release of
CDT with a teleconference next week. I'd also like to hold that telecon 
monthly, if everyone agrees.

As for in-person meetings, I do believe they have
a lot of value, and probably make the most sense when discussing the
design of large new features. In any case, this is something we can 
discuss in the telecon.


Comments and feedback welcome!

The CDT team.