Legacy patches - should I wait?

[Steve Kargl]

On Mon, Jan 21, 2019 at 03:33:14PM +0000, Mark Eggleston wrote:
> I have some patches to support some legacy Fortran features. As I 
> understand it development is now restricted to regression testing and 
> documentation fixes.
> 
> When would it be suitable to submit patches for "new" features?
> 
> I have two that are ready barring last minute checks:
> 
> i. using default widths in formats for f, g and i specifiers
> ii. allowing a blank specifier following a comma, e.g. "(I6,)"
> 

For i, do you mean code like

     write(10,10) x
10   format("(f)")

Fortran has list-directed output, and gfortran already specifies
default widths for the different types and type kind parameters.
The above should be written as

     write(10,*) x

Instead of "when?", I think the question is "why?".  Having worked
on gfortran bugs for 15+ years, I do not recall ever seeing a PR
requesting these (mis)features.  It must be quite uncommon.  What
compilers support these nonconforming constructs?  gfortran currently 
issues errors for ii and (my interpretation of) i.  We should be
actively encouraging users to write standard conforming code.

-- 
Steve