Re: wrapper rpath proposed fix

[Bob Friesenhahn]

On Sun, 28 Nov 2004, Peter O'Gorman wrote:

> Bob Friesenhahn wrote:
>
> > > Assuming that you have verified that it fixes your problem, yes, please 
> > > do.
> >
> >
> > Yes, it does fix this problem and the libtool tests still pass.
>
> Okay, then the 47-peter-pass_new_tests.patch can be reverted in HEAD, this 
> was applied to HEAD as a partial fix for the problem, but was never applied 
> to branch_2_0. Since branch_2_0 now works for you, this patch is superfluous.
>
> I guess?

I assume you are talking about this patch:

2004-10-23  Peter O'Gorman  <address@hidden>

        * config/ltmain.m4sh (func_mode_link): Make Gary's new tests
        pass. When the output wrapper script is not in the current
        directory, we can not use an rpath which is relative to the
        current directory. Use an absolute path instead.

> From the description, it does sound redundant.  Hopefully my patch is 
simpler/better.

If your patch was sufficient to get Gary's new tests working properly 
in *all* situations, then Gary's new tests must be defective.

The biggest concern when dealing with targets in subdirectories are 
compiler/platform specific rules which assume that the .libs directory 
is directly under the current directory (due to simply using $objdir). 
These problems crop up for various C++ compilers that need a place to 
store template instantiations for the library.  They might show up for 
Cygwin/MinGW as well.

I must admit that I have not looked at Gary's new tests to see what 
they are actually testing.  I will do so now.

Bob
======================================
Bob Friesenhahn
address@hidden
http://www.simplesystems.org/users/bfriesen