zack's todo list

[Zack Weinberg]

On Wed, Nov 15, 2000 at 08:00:23AM +0000, Joseph S. Myers wrote:
> On Tue, 14 Nov 2000, Zack Weinberg wrote:
> 
> > static const char a[] = "blah";
> > static const char b[] = "blah";
> >
> > - must a and b be different?  The embedded people would surely like
> >   them to be merged.
> 
> I see nothing to suggest that separately declared objects, or a declared
> object and an unnamed object, can be merged; whereas there is explicit
> wording in 6.4.5p6 to permit string literals to merge, and in 6.5.2.5p8 to
> permit compound literals with const-qualified types to merge with each
> other and with string literals.

Well, I see nothing to suggest they can't, either, but the presence of
explicit wording permitting merging of string literals suggests that
the committee meant for other things not to be mergeable.  (Wanna take
this up with comp.std.c?)  I'll see what I can do.

> > > and a desirable feature would be DESTDIR support;
> >
> > I don't know what this is.
> 
> 	configure --prefix=/usr
> 	...
> 	make install DESTDIR=/some/where
> 
> should install under /some/where/usr (and a packaging system is presumed
> then to handle getting the software into /usr on the system where the
> final install is made).

Ah, I get it.  Can do.

> Once we know why gcc gets the i686-pc-linux-gnu-gcc link and g77 doesn't,
> for example, perhaps it would be a bit clearer when version links should
> be applied.

Well, 'why' is because each Make-lang.in is responsible for installing
its own driver (if any) and they all do it differently.  This is one
of the things I wanted to fix.

> I'd say it should apply for all drivers and any other programs where
> having the right version rather than the latest version is important
> - so not for gcov, for example, since gcov data files have a
> standard format that doesn't depend on the version or architecture.

I think that drivers are the only programs installed outside
$(libsubdir) that fit this description.

zw