Re: removing the 'make install'-->'make all' dependency

[Nicola Pero]

Hi Helge,

I think we'll keep the 'make install' --> 'make all' dependency for
consistency with other GNU packages ... {at least for now :-)}

We'd rather work on fixing the fact that when you type 'make all' a second
time after everything is build, *nothing* must be executed (by *nothing* I
mean *nothing*, there should be no subprocesses except 'make').  At the
moment this is nearly so, but not exactly, at least for bundles,
frameworks and applications, and this is where most of the installation
problems happen.  If my radical medicine of removing the 'install' -->
'all' depedency is not used, then we need to make the build system more
robust by preventing the second 'all' from executing anything (usually the
build fails when it's trying to execute a gnustep-based subprocess).

It has long been a thing I wished to fix anyway, and if 'install' is to go
on depending from 'all', then it becomes a strict requirement that this
fix (which will/might require lots of changes in makefiles and makefile
APIs) is done.

Sorry for not answering the rest of the email (and the other emails), I
think it wouldn't add much new to the discussion if I did, so I'd leave it
here.