Re: A patch for configure

["H . J . Lu" <hjl at lucon dot org>]

On Thu, May 18, 2000 at 09:29:23PM -0300, Alexandre Oliva wrote:
> On May 18, 2000, "H . J . Lu" <hjl@lucon.org> wrote:
> 
> > How about this one?
> 
> It still uses test ... -a ...

I will change it to &&.

> 
> > My goal is people who are not familiar with gcc can build the whole
> > thing with little confusion.
> 
> I doubt people who are not familiar with gcc will have glibc installed
> in a non-standard location :-)

They are in the standard location for cross compile. For me, they
are under /usr/ia64-cygnus-linux/ia64-cygnus-linux. My prefix is
/usr/ia64-cygnus-linux. The only difference is we are using glibc,
not newlib and glibc is not the part of the toolchain.

> 
> Do you know of anybody other than you that has ever encountered the
> problem you describe?  Just curious :-)

It is becoming an item for FAQ for the Trilian project. If you
use anything other than newlib for cross compile, you will see
my point.

> 
> > People like you can always work around this annoying feature without
> > much trouble.
> 
> Indeed.  So can people like you.  That's why I'm still a bit unsure
> about whether this patch should result in a hard error by default.

It should be the hard error by default. That is one reason why I
added -d newlib. Should I add it?

> I'm more inclined to a warning message followed by a `sleep 30', or
> the testing or a variable such as `$enable_target_dir_sanity_checking
> != no', so that someone can skip this test with
> --disable-target-dir-sanity-checking.  What do you think?
> 

I won't mind that. Want a new patch?


H.J.