Re: bootstrap failure, gcc HEAD, --disable-shared, gnatlib-plain

[Arnaud Charlet <charlet at ACT-Europe dot FR>]

> Do we actually know what this is there for?  I was duplicating existing 

I just checked, and it's actually a relic, no longer used.
So it can indeed be removed, and the makefile simplified.

> Well, in the long run, yes.  Some discrepencies are already there, 
> unfortunately.  If the long run plan is to move the Makefile rules into 
> the libada/ directory (as I thought it was; it's probably necessary for 
> proper multilib support), the targets will diverge further again, as 
> well.  If they're *going* to diverge, then....

I suspect it's not strictly necessary for multilib support, although it might
help (not sure about that).

Note that it will help my work of synchronizing the Ada trees if this kind
of change is synchronized with me before hand, and not commited before
having discussed it, because these changes are very tricky (as shown by
the various steps the previous change required), and also because since
we have to keep several trees and branches synchronized, we can't at the
moment use --enable-libada at AdaCore. So basically any change from
ada/Makefile.in/Make-lang.in to libada/Makefile.in can cause merge troubles
if not synchronized properly. I understand these concerns are not strictly
speaking 'FSF concerns', but in the context of helpful cooperation for
having a good Ada tree at the FSF, helping these merges is clearly desirable
and should not be made harder than necessary.

Arno