On Thu, 11 Mar 2010, Derek Gaston wrote:

> On Mar 11, 2010, at 11:13 AM, John Peterson wrote:
>> Hmmm.... Does it need to be a pointer so that we can support the
>> --separate-libmeshout command line option, so that it can be reseated?
>>
>> This is a pretty big change, did I miss a memo somewhere?
>
> Agreed... what is driving this change?

Two things:

We're starting to use libMesh as a more minor subcomponent of programs
incorporating other libraries, and I thought it would be useful for
users to be able to do our tricks (redirect output to per-processor
files, shut up cout on all but processor 0) with the libMesh output
without having to affect cerr/cout of other code.

I'm looking at adding a new trick (sending messages in a "packetized"
form to processor 0 so that they don't get interleaved when printed to
screen) that would affect stream buffering, and I wanted to make it
easy to do that with libMesh debugging output without worrying about
how the behavior change would affect other cout uses.

But I should definitely add option F to my previous list:

F. Revert the whole change, with "what was Roy thinking?" in the svn
log.  ;-)
---
Roy

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to