Re: [Plplot-devel] Status of fortran 95 and my on-going review of logical variables in argument lists for all affected languages

[Arjen M. <arj...@wl...>]

Alan W. Irwin wrote:

>
> I have asked Arjen to change (if it is possible to do so) the way we 
> treat
> call-back functions in our fortran 95 interface and examples so 
> hopefully we
> will see a further fortran change from him in the next day or so.

Callback functions as such are no particular problem - see for instance 
example x19f.

The real problem is the use of an "untyped" argument for these 
functions, if we really
want to be this general. Fortran does not have an equivalent for (void 
*). There are
ways around it, but at a cost. For now I would rather like to 
consolidate the various
flavours we have now under one overloaded name - where the list of 
arguments
determines the actual version to be used.

I can imagine a function/subroutine that takes its specfic data from a 
different
source (a global/module variable for instance). That is easy to add. 
Another solution
is the use of a fixed data type that is general enough for common cases 
- an array
of reals perhaps. The most general case would require some Fortran magic or
a function/subroutine with an unspecified interface. Hm, I do not want to go
that route yet.

Regards,

Arjen