On Sun, 10 Apr 2011, David Knezevic wrote:

> This change seems to have affected ex23. If you put ex23 in "online mode" by 
> setting
> online_mode = true
> in ex23.in, then the RB_sol.gmv that is generated is just zero. Reverting to 
> r4332 restores the old behavior (non-zero RB_sol.gmv).
>
> I guess the change to r4333 affected the writing/reading of RB basis 
> functions in
> RBEvaluation::write_out_basis_functions() and 
> RBEvaluation::read_in_basis_functions() ?

Yup - I'd foolishly assumed that nothing was directly using the
lower-level System::read_* calls without doing a read_headers (and
learning the data file's vector names) first; the library was already
making such an assumption with the cached _additional_data_written
variable so I figured no harm in adding another.  So my implementation
fills a _written_var_indices vector, based on those header contents,
with which to do any later remapping of vectors necessary.

So when rb_sys tries to read vectors without reading the headers,
there are no identified vectors to read, and it doesn't read anything.

I'm not sure I know *what* assumptions I can safely make with regard
to rb_sys, so I'm not sure what the right workaround is.  For
instance, I could restore the old "read vectors in blind sequential
order" behavior whenever _written_var_indices is empty... but then
what happens if you do a full read on on data file and then try to do
a header-less read on another?  It'll break again.

Any ideas?
---
Roy

------------------------------------------------------------------------------
Forrester Wave Report - Recovery time is now measured in hours and minutes
not days. Key insights are discussed in the 2010 Forrester Wave Report as
part of an in-depth evaluation of disaster recovery service providers.
Forrester found the best-in-class provider in terms of services and vision.
Read this report now!  http://p.sf.net/sfu/ibm-webcastpromo
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to