From: Kirk, Benjamin (JSC-EG311) <benjamin.kirk-1@na...> - 2013-04-08 20:10:17
For the past week or so there has been a flurry of activity on a libMesh development branch updating the core data structures in the library to take an optional communicator argument, as a step for allowing e.g. a mesh or set of physics which is represented only on a portion of the available processors during an MPI job. We've tested this with a number of application codes and everything that used to work still does, but this touched some 6,800 lines of code, so be aware if you are following libmesh/master that next time you do an update there will be a big change coming.
The API should be preserved, but as always please report any compilation or runtime issues you may encounter!
-The libMesh Development Team
From: Roy Stogner <roystgnr@ic...> - 2013-04-08 20:17:10
On Mon, 8 Apr 2013, Kirk, Benjamin (JSC-EG311) wrote:
> The API should be preserved
Let's amend that to say "the non-deprecated API should be preserved".
If you were using libMesh::init()/close() instead of constructing a
LibMeshInit object, then the last few years' behavior of "yelling at
you about a deprecated method at initialization time" has been
replaced by "failing to compile at all until you change two lines in
your app code".
From: Roy Stogner <roystgnr@ic...> - 2013-04-08 21:14:21
On Mon, 8 Apr 2013, Roy Stogner wrote:
> On Mon, 8 Apr 2013, Kirk, Benjamin (JSC-EG311) wrote:
>> The API should be preserved
> Let's amend that to say "the non-deprecated API should be preserved".
And let's stress "should". I'm actually seeing a bunch of places
where Ben forgot to uncomment a default argument value... I'll try
and have these all fixed up in a new pull request shortly.
Get latest updates about Open Source Projects, Conferences and News.