On Tue, Feb 26, 2008 at 08:25:25PM -0800, Danek Duvall wrote:
> On Tue, Feb 26, 2008 at 05:47:06PM -0600, Nicolas Williams wrote:
> 
> > >                       What ends up in the runpath here by default, anyway?
> > 
> > /usr/lib -- yeah, no runpath should be needed then, right?
> > 
> > BTW, I needed fix_elf_runpath() because libtool was somehow adding
> > $SRC/lib/libsqlite3/... to the runpath of libtclsqlite3 (but not to
> > libsqlite3).  Don't ask me why; I don't know why.
> 
> Because libtool is a pile.  But you also say that /usr/lib ends up there
> by default, so I'm a bit confused.

You're not kidding.  And I'm confused also.  I think /usr/lib showed up
because it was in LD_OPTIONS (but that didn't override whatever libtool
wanted -- it merged with it).  Libtool is turning out to be a nightmare.

> You might want to ping Stefan Teleman, as he wrestled against libtool for
> many weeks in some of the SFW components.  I think he managed to win, but I
> can't remember now how he did it.  It'd be nice if we didn't have to resort
> to elfedit to make these changes, since we *are* building from source,
> however uncooperative.

OK.  And if we could put together a guide for others who might follow,
then that'd be even better.

> In fact, I'll bet if you patched the build process to include the mapfile
> stuff, they might actually take that upstream.  And even if they didn't,
> maintaining the patch to make that happen might be no worse than
> maintaining the script to elfedit the scoping in after the fact, however
> badass the elfedit-fu might be.

I fully intend to send patches upstream.  I'd like to keep those real
simple though.

Reply via email to