When 4.10 was ... Re: How is the status of 4.8 branch

[Joel Sherrill]

Ralf Corsepius wrote:
> On 08/05/2009 11:06 AM, Leon Pollak wrote:
>   
>> On Wednesday August 5 2009, Ralf Corsepius wrote:
>>     
>>> I strongly recommend not to use 4.8 anymore, but to upgrade to at least
>>> 4.9.
>>>       
>> I am considering an upgrade (from 4.8) too. Which would you recommend? 4.9 or
>> wait till 4.10 (near future?)?
>>     
>
> Difficult question - Answer: It depends.
>
> For "new developments" I'd recommend to follow 4.10, for older works, 
> I'd recommend "upgrading" to 4.9.
>
> BTW: At the moment, there is no ETA for rtems4.10.
>
>   
When it is ready. ;-D

When it comes to predicting release dates there are a few
of driving factors:

(1) when the CVS head has "enough" changes/improvements to
    justify a new release
(2) when new work is done. If someone is working on something
    nice and new and is close, it may be worth waiting.  
(3) Test runs look good on simulators and real hardware feedback
    from community is received.
(4) Tools are in good shape.
(5) when select user projects report they are happy.  The projects
     vary but we like reports from EPICS.
(6) Not in August.  :)

The tools stay in good shape now although a few years ago, the
transition through gcc 3.4 and 4.0 was bumpy.

For 4.10, the focus on testing has meant that the
code on the CVS head stays in good shape.  Chris Johns has built
a daily build infrastructure which publishes a web report.
One of the GSoC students (Santosh) has worked with me
to improve the test coverage.  I am about to make an
announcement with details about the binary test coverage
analysis of RTEMS so won't say a lot.  It is very good though. :)

The "new work is done" is a hard call since something is
always underway which won't make the timeframe.  But for
4.10 there are a handful of projects underway which would
be nice to include and appear to be on track for this fall.
This includes the Summer of Code projects which should be
wrapping up in the next month or so.  Over 1/2 of these projects
already have code in the tree.

I would not be afraid of moving to 4.10 for a project since
(1) it is in good shape and (2) you can then be one of the
user projects providing feedback into when it is time to
cut the release.  The new features shouldn't get in your
way unless you need them.  Then you can help ensure
they are ready for production use.

This was long but there is no magic single rule.  We want
quality releases with as many new features and BSPs as
possible.  We want releases to meet the requirements of
real user projects.  So all of those are balanced when
deciding when the cake is done.

But the CVS head is under continuous testing and integration
along with the new tools.  It stays in good shape.

--joel

> Ralf
>
>
>
>
> _______________________________________________
> rtems-users mailing list
> rtems-users at rtems.org
> http://www.rtems.org/mailman/listinfo/rtems-users
>