Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[cdt-dev] Meeting minutes -- March 5th meeting

Hi,

I've finally found some time to type up the meeting notes from last week's
conf call. Here they are:

Present:

QNX
- Dave Inglis
- Peter Graves
- Sebastien Marineau
- Mikhail Khodjanis
- Judy Green
- Mike Lemke
- Alain Magloire

Redhat
- Phil Muldon
- Andrew Haley
- Rick Mosley
- John Healy

Timesys
- Sam Robb

Rational
- Michel Paquet
- Doug Schaefer
- Sean Evoy
- John Camelon, 
- Andrew Niefer
- Hoda Amer
- Martin Lescuyer

Tensilica
- Chris Songer


1. Update from participants:

QNX -- 6.2.1 Momentics shipped, it is based on CDT 1.0.1.

Rational (Doug) -- New parser code checked in, and build model from
Sam checked in. Rational have posted their plans on the web site; the main
work in the short-term is with the new parser, the model, working copy, and
the build model.

Redhat (John): working on profiler plugin as well as making Eclipse 
compile under gcj

Timesys (Sam): Timestorm 2.0 released, working on 2.1 release. The 2.1
release will be based on CDT 1.0.1


2. CDT 1.1 update
- debug -- work going well, posted list of features is complete. However,
issues were found with gdb 5.3, which forced the implementation of various
workarounds in the CDT to avoid gdb crashes. gdb bugs have been submitted
to the gdb bug tracking system.
- build model status: Sam's builder has been commited, and for 1.1, we
intend
on shipping both the old "standard make builder" as well as the new managed 
make builder. One issue that was brought up was that some companies may
want a way to disable the new builder (esp. those that will ship commercial
products with 1.1). Some straightforward mechanism will be put in place
to achieve that.
- Another request was to have the new parser in 1.1 enabled, with some level
of maturity. In order to accomodate that, we've put the timeframe for the
CDT 1.1
GA as end of April, with the feature freeze a month earlier. This timeframe
was in
line with everyone's requirements.

3. Development activities

CDT Core:

The plan that Rational presented at the last conference call still
stands -- finalized parser and build model for an August timeframe,
while the CDOM would be complete by the end of this year.

A few other points were discussed. The first was the support for K&R
C by the parser. Doug and John thought this would be possible. 

The next discussion involved the indexer -- Alain mentionned that QNX needs
to
find a solution to this rather quickly, due to customer requirements. 
As the debugger work winds down, Alain suggested he will take a look at the
indexer next. The idea of reusing a significant part of the JDT indexer
framework was proposed -- the CDT team will sync up with the JDT team to 
find a way to share the code.

Two more topics were discussed. The first was that QNX will need to do some
changes to the core model, and so will need to coordinate those activities
with Rational closely. Secondly, the class browser was mentioned (this is 
something QNX had suggested they could take on) but no timeframe was set.

Debugger - Sam -- need concept of target model
Mixes C/C++/Java debugging, compiled Java

CDT Debug:

Two topics were discussed. First, Sam reiterated that we need the concept
of a target model. This has been discussed in the past, with Thomas
suggesting
he would lead this. Thomas was not on the call to give an update on this.

Secondly, the long-term direction of the debugger was discussed. The CDT
team
needs to come up with a long-term vision for the CDT debugger, as the
implementation
of features is currently winding down. One of the ideas mentionned was to
tackle
the mixed Java/C/C++ debugging in a single process.


4. Eclipse 2.0 vs. 2.1
The concensus from the participants was that the CDT development should
move to 2.1 as soon as possible. While most agreed that there was no 
compelling argument to move to 2.1 (from a CDT perspective), it was 
thought to be best to stay in line with the current Eclipse development.

It was also agreed that CDT 1.1 would work with 2.1 (and thus would not be
compatible
with 2.0.x). However, it would be good to at least document where in the
code
the incompatibilities reside, in case someone needs to backport CDT 1.1 back
to
Eclipse 2.0.x.


5. Infrastructure:
A couple of quick points were mentionned. First, the bugzilla voting 
has been turned on. Second, we've re-submitted the CVS scripts (to do
auto-commit notification) to the Eclipse infrastructure team.

We also discussed communication, which is becoming more important as
the CDT team grows. It was agreed that a weekly status email from the 
different developers (outlining what they are working on/status) would
be beneficial. 

Finally, it was mentionned that since Doug is commiting the work
of the other Rational contributors, it was important to highlight
the work of the individual contributors. This will help get them 
recognition be voted in as commiters.


6. Documentation and testing

Documentation:  Vishna from Rational is working on user documentation.
Rational is still analysing what needs to be done and what areas
need to be documented. In the short-term, they are also looking
at building a tutorial on CDT, which should greatly help new users.

Testing: The JUnit testsuite continues growing. While there's still a long
way to go to cover all existing features with testcases, the new core
features
have associated with them testcases, which is a great step forward!

Automated UI testing: After the discussions from last month, QNX has
proposed
the we use Winrunner as the UI automation tool. Rational will do some
evaluation of this and come back to the list with their preference; 
Chris mentionned that Tensilica was already evaluating tools for 
UI testing (Winrunner was one of the ones under eval): 
Chris will come back to the list with the conclusions of his test team
in a couple of weeks.


7. Varia

CPPUnit was discussed. Work is already underway at Rational on a CPPUnit
integration
with Eclipse, so it is important to find a home for this. Michel proposed
that 
this be rolled under the Hyades project. I will follow-up with JohnD and the
Eclipse
folks to resolve this as quickly as possible.


That's all I had for notes. If I've forgotten anything or mis-quoted anyone,
please let me know.

Thanks,

Sebastien


--------------------
Sebastien Marineau
QNX Software Systems
(613) 271-9336
sebastien@xxxxxxx 


Back to the top