On Friday 12 June 2009 Jim Gallagher wrote:
> The patch works well, thank very much, it's a life saver.
>
> I changed
> -Wl,+h$(SPACE)-Wl,$(<[-1]:D=) to
> -Wl,+h$(<[-1]:D=).
> Looks a little cleaner and works the same.
I have finally checked this in in SVN.
> So, everything looks good. For my case, I will probably add
> +nodefaultrpath, which will
> prevent the embedding of the paths supplied with -L. Those paths are
> needed at link time,
> but they are not appropriate for our runtime configuration.
Aha, I see.
> To shift gears a bit, I have been following the mailing list for a
> while, and I agree
> with your (the maintainers) philosophy, that boost-build should not
> surprise users
> and should do what is usually the right thing. I think this is a case
> that, at least
> for HP-UX and Linux systems, and maybe for Unix in general, that
> boost-build is doing
> something that is surprising and not what most developers would want.
> I suggest that
> the default boost-build link lines for Unix systems should use the -L
> and -l path
> and lib specifiers.
Well, -L/-l is less precise, while specifying full name guaranteed that linker
will use the specified file.
- Volodya
_______________________________________________
Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost-build
