Re: [RFC] "C" header options

[Benjamin Kosnik <bkoz at redhat dot com>]

> I too have reviewed Stephen's work in this area, as last posted by
> Benjamin (although I haven't added it to any source trees).  I like
> the idea of moving to "everything to be installed as a header is
> staged/built".

Yeah. The patch has evolved since I last posted it.

It's too big to go in at this point. 

Sorry. 

First thing for gcc-3.1.

I've been able to get some of the make clean issues resolved with:

# By adding these files here, automake will remove them for 'make clean'
CLEANFILES = \
	stamp-backward stamp-bits stamp-c_base stamp-ext stamp-libio \
	stamp-std stamp-target 

But this only forces the re-copy, not the removal of copied files. (This 
is important, because make may copy a smaller number of files.)

So, the dependency tracking stuff is not quite up to snuff right now. 
(Of course, this is also an issue with what's in CVS...but I digress.) Other 
than that, it's nice and more importantly, the way development should go.
One of the most interesting things for me was to understand how somebody 
else viewed the process of making headers, and the resulting 
make/configure changes. I hope that this will eventually make the 
libstdc++ configure/build process more transparent.

-benjamin