Making GDM not support fallback may have a more serious effect than making this decision for the rest of the GNOME desktop.  Modules that are used to ensure system security, like GDM, particularly suffer from any lack of cooperative effort caused by forking.  When forking happens, managing things like security
bug fixes becomes more complicated.

For example, if you GDM stops supporting fallback, then you make users and 
distros that need to support older hardware switch to some other display manager
or get stuck supporting an older version of GDM, effectively forking.  

I should think that encouraging the forking of security related modules like GDM would be much less desirable than encouraging forking of the rest of the desktop.  So, please make this kind of decision with a more full awareness of what you are really deciding.

well note, the shell based greeter works on older hardware too, because of software rendering.

One idea is, we could move the fallback greeter out of tree (just as the shell greeter is out of tree).  That way the core daemon code is still unforked, but anyone needing to maintain the fallback greeter could continue to in the other module.

Here's what we agreed to on d-d-l, considering that it is quite late in the cycle: we'll just add a --enable-fallback-greeter configure option, the code will remain in the gdm module for 3.8

comment 3 already landed:

https://git.gnome.org/browse/gdm/commit/?id=acc931c762b5510103a7f49cf3074c1228700cb8

So i've expunged the fallback greeter from git now.

If anyone wants to resurrect it as an independent module in git,
feel free.

By the way, gdm 3.8.3 dropped the fallback greeter. I guess this was caused by not branching gdm for 3.8.

ah, that's accidental. I'll do a 3.8.4 that adds it back.

i called it 3.8.3.1