[patch, fortran] PR22539 implement FSEEK intrinsic

[Daniel Franke]

On Thursday 03 May 2007 23:22:05 Brooks Moses wrote:
> > With all the changes in place and no regressions on i686-pc-linux-gnu, ok
> > for trunk?
>
> I do have one other question, which I apologize for neglecting to
> mention earlier: you mentioned, at some point, handling various kinds of
> the STATUS variable by some form of conversion, if I'm remembering
> correctly.  I don't see any of that in the patch, though, and it looks
> like you're constructing a call that passes along the STATUS argument
> with whatever kind it happens to be, and then fseek_sub expects an int*.

I mentioned, that the conversion still needs to be done, not that it is 
available yet :)

That is the very same problem as other intrinsics have, e.g. EXIT, SIGNAL, you 
name it. Most, if not all, that feature a status variable. I still plan to 
find a general fix for the problem and to get rid of the related foo_i[1248] 
functions in the library by arranging for appropiate casting before and after 
the call, respectively. I'm still not sure on how to accomplish this as I am 
sort of lost with the creation of gimple, but eventually I'll figure it out. 
Thus, I concluded that there is no need to water-tight FSEEK in this respect 
as there are other functions/subroutines -- that are used more often -- that 
still need to be fixed. I.e. I plan to get back to it as soon as I know how 
to create the appropriate casts. Is this acceptable for now?