On Thu, Dec 28, 2000 at 10:24:46AM -0500, Brian S. Julin wrote:
>
> On Thu, 28 Dec 2000, Nicolas Souchu wrote:
> > You mean the implementation of the extension-system based on dl functions?
>
> IIRC, there's some work to be done in being more sophisticated about which
> extension sublibs get loaded in which order. Marcus may have other agenda
> items.
>
> (Marcus -- Something I noted is that edemotest with the X target segfaults
> on exit.)
>
> > By the way, has one of you experienced ggi on an embedded system which does
> > not have dlopen?
>
> Marcus made some recent changes to make it easier to statically link
> LibGGI and some targets/extensions, but there is no build option
> to do it yet. It involves substituting the functions that do the
> dlopen with simple table lookups.
Would we need the heavy libtool environment then? Wouldn't it be wise to have
a separate distribution for embedded systems? Without configure, libtool, but
with sophisticated makefiles and a more limited list of systems on which it WORKS.
What makes the power of GGI today is more it's distribution package, implementation
or interfaces and design? The last I think.
--
[EMAIL PROTECTED]
Alc�ve - Open Source Software Engineer - http://www.alcove.fr
