(j3.2006) Fwd: BOZ incompatiability in Fortran standards

[Van Snyder]

On Fri, 2017-10-06 at 23:02 +0000, Bill Long wrote:
> But if you really have an array initialization that takes 600 lines of
> code, I?d first suspect that the code design needs some work. 

Tell me what's wrong with the design of the attached module, for which
the array initializer takes 771 DATA statements, even with 7708 lines of
named constant definitions that provide named constants for the DATA
statements.

If we had changed the maximum number of lines in a statement to 9999, as
I suggested during the development of F03, the variable initialized
using DATA statements could have been a named constant without the need
for an additional level of declarations of sub-arrays.

But even at 9999 it would have been a close deal, because this case has
7708 lines of constant value definitions, although it would have fit in
3851 lines because I wouldn't have needed to declare a named constant
for each array element.  But that's still too many for what we allow
now.

So don't obsolesce DATA until the statement limit is ridiculously large,
and 9999 might not be ridiculous enough.

Now that we've decided we really really must have the goofy "A is sort
of like B except for X, Y, and Z" for locality specs, which is far
goofier than the "A is sort of like B" for statement function dummy
arguments, can we un-obsolesce statement function definitions?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Big_Const.f90
Type: text/x-fortran
Size: 665489 bytes
Desc: not available
URL: <http://mailman.j3-fortran.org/pipermail/j3/attachments/20171006/fb785fa1/attachment-0001.bin>