Re: a history question
From: Richard E Maine (nospam_at_see.signature)
Date: 09/29/04
- Next message: Ian Chivers: "Re: a history question"
- Previous message: *** Hendrickson: "Re: KIND related issues on NAGware f95 v 4.1"
- In reply to: Rich Townsend: "Re: a history question"
- Next in thread: Arjen Markus: "Re: a history question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 29 Sep 2004 08:40:29 -0700
Rich Townsend <rhdt@barVOIDtol.udel.edu> writes:
> How were you able to actually get at the bits of a pointer? I thought
> this wasn't possible in F90/F95 --- transfer on a POINTER type acts on
> the target of that type, yes?
Yes, but the "usual" trick works. I alluded to it briefly, but didn't
spell it out. Make a derived type that looks like
type whatever_ptr_type
type(whatever_type), pointer :: ptr
end type
This kind of wrapper type allows you to do several things that you
can't directly do with a pointer. The most common case is wanting
arrays of pointers (as distinct from pointers to arrays). For the
current discussion, it allows you to do
type(whatever_ptr_type) :: whatever_ptr
...
transfer(whatever_ptr,...)
which does the trick.
But don't try this particular variant at home, particularly if your
compiler does garbage collection. :-(
-- Richard Maine | Good judgment comes from experience; email: my first.last at org.domain | experience comes from bad judgment. org: nasa, domain: gov | -- Mark Twain
- Next message: Ian Chivers: "Re: a history question"
- Previous message: *** Hendrickson: "Re: KIND related issues on NAGware f95 v 4.1"
- In reply to: Rich Townsend: "Re: a history question"
- Next in thread: Arjen Markus: "Re: a history question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]