Re: Artificial front-end variables?

[Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>]

Tobias Schlüter wrote:
> Anyway, I'll hope to have this in a shape where I can apply it to some 
> simple cases, and see if I can use it to fix some of the aforementioned 
> problems tomorrow.

I have the code in place, but I couldn't yet fix any problems with it, I 
 thought it would be easier to insert code, while the compiler is 
looking at expressions.

The problem I'm seeing is that I have to generate code for the 
assignments, but the only place where I can do this without introducing 
an additional pass, is in resolve.c in resolve_code, while the code list 
is traversed, and the subexpressions are resolved, which doesn't seem 
clean.

I also looked at implementing statement functions in the frontend using 
temporaries (pr14454 seems easy enough to fix), but that would also 
involve some magic, the current implementation is already in 
gfc_se-world, so it doesn't seem adaptable to frontend temporaries.

Steven, did you habe a plan on how to use temporaries for implementing 
SELECT? Maybe that procedure would be adaptable to the problems I wanted 
to tackle.

- Tobi