All,
after toiling for three days to get through our release checklist, there are 
three issues worth bringing up:

- Test the release and let us know if you find things that don't work. This is 
the first release that includes the Threading Building Blocks and I'm sure 
there are issues with building it and building with it that we have missed. 
As always, there will also be bugs when you ship 500,000 lines of code, so 
let us know if you find something that's broken.

- Timo Heister, Martin Kronbichler and I have been working on the 
infrastructure to make possible computations on meshes that are too big to be 
stored on each MPI process (think: 1 billion cells, several billion 
unknowns). We've got this all pretty much finished and it scales to the 
biggest machines we've had access to so far with 4,096 processors. It depends 
on a library that isn't publically available yet, but we'll have this code 
merged whenever it becomes available.

- There are also a couple of tutorial programs that are in the making. Martin 
Kronbichler and Katharina Kormann have a program that implements multigrid 
methods with matrix free algorithms. There will also, at one point or 
another, the long delayed step-26 with complicated boundaries, and step-32 
and step-40 will cover the distributed meshes facilities discussed above.

If you think you have some functionality that would be cool to have in 
deal.II, or if you'd like to work with us on something that could go into 
deal.II, now is the time to let us know about it (on or off the mailing 
list). You've seen the number of people who have contributed to 6.3.0 -- take 
it as an indicator that we're more than happy to work with people to get 
stuff in, and that we like it if others contribute.

In this regard: I've made the mistake of announcing too late that 6.3 is 
imminent and it turned out that a couple people still had stuff lying around 
that in the end didn't make it into the release because we couldn't test it 
sufficiently. That's my fault but I'll take the opportunity to say that at 
the beginning of a release cycle is the time to get stuff merged with plenty 
of time to get it tested, iterate on issues and have it ready for the next 
release.

Best
 Wolfgang

-------------------------------------------------------------------------
Wolfgang Bangerth                email:            [email protected]
                                 www: http://www.math.tamu.edu/~bangerth/

_______________________________________________
dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii

Reply via email to