On 14.04.21 09:57, Jonathan Wakely wrote:
On Wed, 14 Apr 2021 at 08:46, Thomas Koenig via Gcc <gcc@gcc.gnu.org> wrote:
There is no discussion at the moment. Most people on the fortran
mailing list do not follow gcc. I know of at least two contributors
(myself incluced) who would in all probability stop contributing
in that case.
Do you mind if I ask why?
(I totally understand that you'd rather not have this topic spill over
onto the gfortran list, so I'm only asking why you'd stop contributing
if there were two active forks of GCC, not anybody else).
Because I am not willing to donate my time and effort for a doomed
project, and if that split happens, I consider gfortran doomed
for good.
Let's look at
- All gfortran developers stay on the FSF branch.
Bug fixing goes on as usual, the other branch picks up whatever
it wants. This I could see as working, sort
of. If the FSF and the other branch diverge in their middle
end interface, or if the other branch decides not to port something,
this is bit rot that the maintainers of the other branch
would have to deal with. So, gfortran bitrots on the new
branch, basically. The question then is if the FSF branch
will still be the relevant one the future. If not, gfortran
will then die a lingering death.
- gfortran developers try to work on both branches, or have
one primary branch and one other branch. Dealing with two
versions is far too much for our resources, we can hardly
keep up with one. This is a recipe for disaster, and
I will not spend my volunteer time on this.
- Some gfortran developers decide to move to the other branch,
cross-porting fixes if necessary. This will also lead
to fewer resources of a project that has already too few,
and is not sustainable.
- All gfortran developers move to the new branch. This will not
happen, I can guarantee you that.
- Somebody decides that hiring a couple of professional programmers
working full-time keeping the project alive on both sides.
That has not happened in decades (gfortran has always been
mostly volunteer-driven), and I consider that extremely
unlikely.
So, let me modify my statement: It only makes sense for me to
continue working on gfortran if the branch fails. In that case,
the maxim usually employed by pharmaceutical R&D applies:
"Fail early, fail cheap", and the cheapest way to fail is
never to start.
