F2003's IOMSG feature and Janne's FLUSH patch

[Richard E Maine]

On Jul 30, 2005, at 1:21 PM, François-Xavier Coudert wrote:

>> I've finally found some time to review Janne's implementation
>> of the F2003 flush command.
>> (1) Should we implement IOMSG before committing Janne's patch?
>>
>> (2) Can I commit Janne's patch with an appropriate TODO comment?
>
> I'd go for (2). F2003 is nice, but seeing the rather large number of
> bugs on gfortran's plate, I don't think F2003 features are high
> priority.

I'd agree with the following additional justification. I view iomsg as 
a separate feature from FLUSH. A very nice feature, I might add, but 
orthogonal to flush. FLUSH isn't even one of the more likely places for 
it to be wanted. It is just in flush for consistency. If you put it 
only in flush, you'll have an inconsistency... and a darned annoying 
one because you'll have it in one of the least useful places and not in 
the more useful ones. Now adding iomsg to OPEN, READ, and WRITE - that 
would be useful. Just last week I was helping yet another person deal 
with "what does error status 175 from OPEN mean?". It's a PITA that the 
only way to get a decent error message from OPEN is to give up on error 
handling and let the system abort the program. I'm not asking for you 
to do iomsg now, as I agree with needing to get f95 working as a higher 
priority. I'm just pointing out what I would see as a reaction to 
putting iomsg in one of the least useful places for it.

There are lots of features that interact in some way or other - you 
can't reasonably say that you won't implement one feature until you 
also do everything it interacts with. That way lies doing nothing until 
you have the whole language.

-- 
Richard Maine                |  Good judgment comes from experience;
Richard.Maine@nasa.gov       |  experience comes from bad judgment.
                             |        -- Mark Twain