(j3.2006) (SC22WG5.4262) [ukfortran] wg5

[Malcolm Cohen]

> Here is a draft revision of my summary of the new features of F2008. Would
> anyone like to comment before I put it on the WG5 site?

Your example of something exit can do that cannot be expressed in F2003 could be 
written as a single statement in F90.  That's not very convincing.  You could 
just tone down the claims about expressiveness.

(Actually, I generally find the advertising to be a bit overstated and 
unnecessary, but it's your paper not mine.)

5.5 example is invalid except in a module or main program.

6.2 explanation of when one had to explicitly put the bounds in an array 
allocation in F2003 is confusing.  One always had to explicitly put them, 
whether one wanted to copy another array's bounds or not.

10.2 is wrongly explained.

10.6 is missing the unallocated allocatable case.

The example in 10.8 isn't what we were trying to achieve, and illustrates some 
oversights in the constraints (an interp request will be forthcoming).  A PRNG 
function would be a better example.

I disagree with Van's comment on 10.9 (because sharing data doesn't require the 
data to be stored in module variables except when they have the SAVE attribute, 
in which case it won't be thread-safe anyway).

Cheers,
-- 
................................Malcolm Cohen, Nihon NAG, Tokyo.