Re: [Libmesh-users] EquationSystems read() and init()

[Jens L. E. <jle...@gm...>]

Thanks; I'm happy with the way I do things now but at the very least 
I'll provide a misc example.

How do I contribute that?

Jens

On 01/26/2013 09:17 AM, Roy Stogner wrote:
>
> On Fri, 25 Jan 2013, Jens Lohne Eftang wrote:
>
>> So I want to have an EquationSystems that holds two systems, one that is
>> read from file, and one that is not.
>>
>> I tried a couple of things:
>>
>> Read system from file, then adding another system, then init()
>>
>> and
>>
>> add a system, then init(), and then read from file.
>>
>> None of this works and seems to, when I print_info(), duplicate the
>> number of dofs in whatever system is inited twice. Also, if I add a
>> vector to this system the size of that vector is twice as large as it
>> should be.
>
>> What does seem to work is to first add a system, and then read from
>> file, and not call init() at all. Is this the right way to do this? I
>> guess it makes sense since read() also initializes ...
>
> I'd have said that the right ways to do things would be either the way
> you ended up doing things, or read(), then add_system(), then
> reinit().  I highly doubt we support that latter option currently, but
> we've changed things in the past to add similar support for
> adding-vectors-after-init, adding-matrices-after-init...  I don't have
> time to add similar support myself at the moment but I'd be willing to
> help if you wanted to work on a patch.
>
> On the other hand, if the way you're doing things now works then that
> should be fine.  I'd still suggest contributing a short miscellaneous
> example program that exercises that functionality, though; it's merely
> good luck that it's working now and you might not want to rely on
> continued good luck alone to prevent regressions.
> ---
> Roy