[J3] gfortran funding (was: Niklaus Wirth)

[Thomas König]

Hi Tom,

> I can only speak on the gov’t side.   If there was a simple way that I 
> could “pay” $100 for a gfortran “license” that would be great and 
> (relatively) simple.  (Don’t ask about the amount of paperwork it 
> requires just to make a $100 purchase on a gov’t credit card under 
> “simplified acquisition”.)   But what is being described here is a 
> _/service/_.   And setting up a contract for service is an entirely 
> different game and not even worth the effort until it is >> $1k.

I can imagine (but I am lucky enough to work in industry and have
very little contact with

> In principle, I could go to my management and tell them that we need to 
> spend say $10k-$20k to get some crucial feature or bugfix from a 
> gfortran developer.   They probably would approve.   OK.  So now we need 
> to find a vendor that can provide this service.    I cannot just send 
> the money to a known gfortran developer.  They have to have a legal 
> entity registered to do acquisitions with the US Gov’t.

Last year, there was some discussion about getting funding from the
German government's Sovereign Tech Fund, which is set up for supporting 
crucial open-source projects (https://www.sovereigntechfund.de/).

An application was made, but it was not even considered, apparently
because there were too many applications coming in.  While this is
not officially dead, the likelihood of getting funds that way is very
low.

The project plan includes:

- Updating the shared-memory coarry branch to current gcc trunk
- Update gfortran's coarrays so (basically) allocatable components
   of coarrays work (ABI change required)
- Enhance the support for teams and failed images in coarrays to a
   level where it gets usable
- Enhance standard compliance from Fortran 2003 onwards (PDTs,
   finalization)
- Ensure maintainability of gfortran by cleaning up/refactoring internal
   APIs including the scalarizer.

Project coordinator was to have been Andre Vehreschild of Badger
Systems; I have copied him in. I don't know if they are registers
with the US government, though.

So, if anybody has any possibilities for gfortran funding now the STF
has effectively fallen through, that would be very interesting (and
could advance the compiler by quite a lot).

> But since 
> we know DOE is funding some such work, there must be one-or-more such 
> entities, and this could be done.  And then it takes time and patience 
> to dot all the eye’s.   Would I still care about the feature/bug in 
> question that many months later?   Or would I have been forced to find 
> acceptable workarounds by then?

There is also the time frame of a fix to consider, if you would be willing
to wait for a next major release, which come out yearly.  Alternatively,
it is also possible to backport fixes to stable branches, if they are
considered important and low-risk.

> Sigh.  (You may notice that I’m a bit burned-out by NASA procurement 
> processes.)

I sympathise.

Best regards

	Thomas