Re: More info on libldtl failure

[Nick Hudson]

On Thu, 1 Feb 2001 16:52:10 +0100 (MET), Michael Matz wrote:

>  Hi,
>  
>  On Tue, 30 Jan 2001, Nick Hudson wrote:
>  > >  Hmmm.  I think I can see why this might be... the entire deplib
chain
>  > >  is followed and loaded for each lt_dlopen()ed module.  Perhaps I
need
>  > >  to cache previous successes and failures to avoid repetition?  Or
>  > >  maybe I just need to provide an option to ignore deplibs (and
emulate
>  > >  the old behaviour that served us so well before)?
>  > 
>  > Gary,
>  > 
>  > I believe your analysis is correct. This libltdl "feature" is causing a
slow
>  > behaviour for my KDE2 packages for NetBSD. I've been playing with the
idea
>  > that the deplibs should be ignored as well as they should be "auto"
loaded
>  > by the OS
>  
>  KDE's copy of libltdl doesn't follow the deplib-chain. In former times I
>  thought about activating it (before it was in libtools libltdl), but as
it
>  exhibits quadratic behaviour I didn't. And I was to lazy to implement the
>  proper solution (a cache of already loaded .la files), so it still relies
>  on the OS.  What version of libltdl are you using with KDE2?

A recent snapshot (23 Dec/22 Jan) of multi-language branch. See
http://www.nthcliff.demon.co.uk/NetBSD/KDE2/index.html. Since this page was
last updated I've started looking at 2.1beta which has more problems for
NetBSD ;-( I digress...

I don't understand what the benefit of following the deplib-chain is? I
guess its to support some missing functionality in some platforms. If so why
can't this be enabled/disabled at configure/build time?

Nick
--
aka address@hidden, address@hidden





_______________________________________________________
 Get 100% private, FREE email for life from Excite UK
 Visit http://inbox.excite.co.uk/