Dummy pencils
Brought to you by:
tobson
Originally created by: anders.j...@gmail.com
Originally owned by: anders.j...@gmail.com
The nomodules must provide certain pencils that are requested by other
modules. These all take up space in the pencil case. However, it should be
possible to have pencil pointers so that dummy pencils can simply point to
an array of size (nx) with just zeros.
View and moderate all "tickets Discussion" comments posted by this user
Mark all as spam, and block user from posting to "Tickets"
Originally posted by: wdobler@gmail.com
Can you give an example where we legitimately expect a no-module to provide dummy
pencils? To me, this sounds like a design flaw.
If we really need this, the array should most probably not contain zeros, but
IMPOSSIBLE values (NaN would be even better, but cannot portably be created in F90).
View and moderate all "tickets Discussion" comments posted by this user
Mark all as spam, and block user from posting to "Tickets"
Originally posted by: anders.j...@gmail.com
The problem is that the code needs to compile, even when moduleA uses pencils from
nomoduleB. See e.g. the header of nodensity.f90. In this case it actually makes sense
to set p%rho=rho0.
View and moderate all "tickets Discussion" comments posted by this user
Mark all as spam, and block user from posting to "Tickets"
Originally posted by: wdobler@gmail.com
If nodensity.f90 needs to provide rho0 in pencils, then those cannot be shared with
other no- modules.
> The problem is that the code needs to compile, even when moduleA uses pencils
> from nomoduleB.
How come this problem hasn't shown up until recently?
View and moderate all "tickets Discussion" comments posted by this user
Mark all as spam, and block user from posting to "Tickets"
Originally posted by: anders.j...@gmail.com
I don't see this as being inconsistent. If module A needs pencil p%xyz, then some
other module must provide this pencil. That will often be a no-module that simply
returns an empty or constant pencil.