This is the mail archive of the
mailing list for the GCC project.
Re: [Patch, fortran] PR23060 - %VAL, %LOC and %REF
- From: Brooks Moses <brooks dot moses at codesourcery dot com>
- To: gcc-patches at gcc dot gnu dot org
- Cc: fortran at gcc dot gnu dot org
- Date: Fri, 09 Mar 2007 01:21:59 -0800
- Subject: Re: [Patch, fortran] PR23060 - %VAL, %LOC and %REF
- References: <45954E3F.email@example.com>
I was going through old mailing-list stuff, and came across this:
Paul Thomas wrote:
This patch implements the argument list "functions" %VAL, %LOC and %REF.
It is straightforward upto the following remarks:
(i) It was not evident how to implement these so-called functions
because their context is so specific. Rather than implementing
intrinsic functions, I have chosen to signal them using the keyword
name field of the actual_arglist structure. This does not lead to any
problems because these relics commence with '%'. I thought that it
would be too cumbersome to make this field a union of two names.
(ii) I have chosen to dodge the CHARACTER question by implementing
these functions as for Digital Fortran; ie. I have not allowed
(iii) Unlike Digital Fortran, complex arguments are permitted in %VAL.
(iv) The tests are a minimum to check the basic functionailty.
Did this ever get documented in the "Extensions" part of gfortran.texi?
It's nice to implement new features and all, but they don't "really"
exist unless people outside the development cabal can use them, and that
means documenting them. :)