[Trilinos-Users] Thyra/Epetra adapters

[Kevin Long]

Ross,

Will it be possible to derive subclasses from the Thyra/Tpetra adapters? I ask 
because some Thyra classes (DefaultSpmdVectorSpace, for example) have private 
constructors making derivation impossible (except through deprecated 
constructors which may go away). 

I know this is because you favor the decoupling of object creation and onject 
initialization, but the lack of public (or even protected) constructors does 
make things inconvenient for people who would like to use Thyra's adapter 
support classes for interoperability but then build a different user 
interface on them.  

Regards,

Kevin





On Friday 25 June 2010 01:33:19 pm Bartlett, Roscoe A wrote:
> Micheal,
>
> The Thyra/Epetra adapters are not all classic ADAPTER classes.  Since
> Epetra_Vector and Epetra_MultiVector are just thin wrappers around raw
> arrays of data, the Thyra adapter code actually just uses
> Thyra::DefaultSpm[Multi]Vector objects to provide a Thyra-compatible view
> of Epetra objects.  There are obvious advantages and disadvantages to this
> approach (that I will not go into here).  Just focus on the conversion
> functions to and from Thyra and Epetra objects as shown in the examples and
> you should be fine.
>
> In the case of the new Thyra/Tpetra adapters (to be released in Trilinos
> 10.4 shortly), we went with all ready ADATPER subclasses because Tpetra is
> more complex and you can't assume that all of the data is in main memory.
>
> Let me know if you have problems with these conversion functions.
>
> - Ross
>
> > -----Original Message-----
> > From: Michael Buksas [mailto:mwbuksas at lanl.gov]
> > Sent: Friday, June 25, 2010 12:27 PM
> > To: Bartlett, Roscoe A
> > Cc: Salinger, Andrew; trilinos-users at software.sandia.gov
> > Subject: Re: [Trilinos-Users] Thyra/Epetra adapters
> >
> > Ross, Andrew,
> >
> > Thanks for the links and examples. That should get me started.
> >
> > The Thyra documentation appears to skip the Epetra adapter classes
> > entirely, but not EpetraExt. For example, Thyra::EpetraLinearOp is in
> > the Thyra class list, but Thyra::EpetraVector is not.
> >
> > > Note: We have no automated testing of our HTML pagers so it is nearly
> >
> > impossible for use to catch all of these errors.  There are tools for
> > doing this type of testing automatically but we don't have the
> > resources to get all of this done right now (hopefully at some point we
> > will).
> >
> > No worries there. Given the scope of the documentation, I'm happy it
> > took me this long to find a problem.
> >
> > Mike
> >
> > > Thanks,
> > >
> > > - Ross
> > >
> > >> -----Original Message-----
> > >> From: trilinos-users-bounces at software.sandia.gov [mailto:trilinos-
> > >> users-bounces at software.sandia.gov] On Behalf Of Michael Buksas
> > >> Sent: Friday, June 25, 2010 11:29 AM
> > >> To: trilinos-users at software.sandia.gov
> > >> Subject: [Trilinos-Users] Thyra/Epetra adapters
> > >>
> > >> Hi all,
> > >>
> > >> Where can I find documentation of the Epetra to Thyra adapters?
> > >>
> > >> Various parts of the doxygen documentation (e.g. the Epetra main
> >
> > page)
> >
> > >> have links to:
> >
> > http://trilinos.sandia.gov/packages/docs/r10.2/packages/epetra/thyra/do
> >
> > >> c/html/index.html,
> > >> but this appears to not exist.
> > >>
> > >> Thanks!
> > >>
> > >> Mike
> > >>
> > >> --
> > >> Michael W. Buksas
> > >> CCS-7: Applied Computer Science
> > >> Los Alamos National Laboratory
> > >> (505) 667-7580
> > >>
> > >> _______________________________________________
> > >> Trilinos-Users mailing list
> > >> Trilinos-Users at software.sandia.gov
> > >> http://software.sandia.gov/mailman/listinfo/trilinos-users
> >
> > --
> > Michael W. Buksas
> > CCS-7: Applied Computer Science
> > Los Alamos National Laboratory
> > (505) 667-7580
>
> _______________________________________________
> Trilinos-Users mailing list
> Trilinos-Users at software.sandia.gov
> http://software.sandia.gov/mailman/listinfo/trilinos-users



-- 
------------------------------------------------------
Kevin Long
Associate Professor
Department of Mathematics and Statistics
Texas Tech University
Lubbock, TX 
------------------------------------------------------