(j3.2006) (SC22WG5.4637) AW: Part 2 of the standard

[Bader, Reinhold]

On Walt's web site there exists a link to a test suite by Lawrie:

http://www.fortran.com/vst_95a.zip

HTH
Reinhold


> -----Urspr?ngliche Nachricht-----
> Von: owner-sc22wg5 at open-std.org [mailto:owner-sc22wg5 at open-std.org] Im
> Auftrag von John Reid
> Gesendet: Dienstag, 13. M?rz 2012 21:04
> An: WG5
> Betreff: (SC22WG5.4635) Part 2 of the standard
> 
> WG5,
> 
> I promised at Garching to look into the conversion to Fortran 2003 of
> the procedures of the varying length strings module that are not
> included in Part 1 of the Standard. These are the I/O procedures GET,
> PUT and PUT_LINE and the substring manipulation procedures EXTRACT,
> INSERT, REMOVE, REPLACE, and INSERT.
> 
> I have made a new module that compiles under the Nag compiler but I have
> tested it only with new versions of the two example programs that are in
> Part 2.
> 
> Most occurrences of
>     type(VARYING_STRING)
> were replaced by
>     CHARACTER(LEN=:),ALLOCATABLE
> but
>     type(VARYING_STRING),INTENT(IN)
> was replaced by
>     CHARACTER(LEN=*),INTENT(IN)
> This resulted in arguments that were previously documented as either of
> type CHARACTER or VARYING_STRING become only of type CHARACTER. The
> number of specific procedures reduced from 40 to 15.
> 
> There were obvious mechanical changes needed within the procedures.
> 
> The subroutine SPLIT cannot be elemental because of having an
> allocatable argument. Note that all the elements of a character array
> have to have the same character length, so this restriction really is
> needed.
> 
> A similar problem occurs in the vocabulary test, where an array holds
> words of varying length. I got round this by using a linked list.
> 
> This has taken me a day and a bit of work. To get it all tested and
> documented will be quite a bit more work, probably at least a week. How
> valuable will this be?
> 
> Lawrie, do you have a view? Can you help? For instance, do you have a
> test deck for the old code?
> 
> Cheers,
> 
> John.