| Richard E Maine 2005-05-14, 4:05 pm |
| In article <d62mfp$4ug$1@meiner.onlinehome.de>,
Thomas Koenig <Thomas.Koenig@online.de> wrote:
> Is it OK to have a namelist in block data
No... at least by my interpretation. I do see a slight possible
contradiction in the standard now that I check carefully, but my
interpretation of the contradiction would be that the "no" is more
explicit than the "yes".
The "yes" side is supported by table 2.2 in the standard, which says
"yes" to allowing "Misc. declarations" in block data. If you check the
table footnote definition, "misc declarations" includes specification
statements, which (per R214) includes namelist statements.
The "no" side is supported by the second constraint in 11.4, which says
"A block data specification part may contain only...[an explicit list
that doesn't include namelist statements]".
That "no" sounds pretty explicit to me. Indeed, because it is a
constraint, a compiler is required to be able to diagnose the error. So
I'd suggest a bug report to the vendor that accepted it (unless there is
at least an option to diagnose it).
I think I'd interpret the "yes" as saying that some of the specification
statements are allowed, but not necessarily all of them. That is an
interpretation on my part (and thus unofficial), I admit, but that's the
only interpretation that I can see as being consistent.
--
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
|