[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openzaurus-users
Subject:    Re: [Openzaurus-users] Re: ssh-client on OZ 5600 libcrypto.so.0.9.7 error
From:       Chris Larson <kergoth () handhelds ! org>
Date:       2003-11-19 22:14:02
[Download RAW message or body]

* Rick Schultz (bloodyvikings@sbcglobal.net) wrote:
> On Wed, Nov 19, 2003 at 12:05:15PM -0600, Chris Larson wrote:
> > I think you're misunderstanding how this works.
> Undoubtedly.  I will now attempt to prove just how poorly I understand
> this...
> 
> > ipkg-link knows about the linking to root, but doesn't have intimate
> > knowledge of what files *should* have been on the card, so can't make
> > that decision.
> 
> I take it, then, that the pkg.list file only contains files that
> actually made it, and not all the files that were supposed to have been
> created?  In either case, if the .ipk file is still around when
> ipkg-link is invoked, that information should be obtainable, right?

Um, ipkg doesnt cache ipk files.  Say I install one from a feed, it
downloads it, installs it, removes it, then we call ipkg-link.  Boom.
Also, the .list contains a list of files, not where they should be
pointing.  It doesnt have sufficient information to do what needs to be
done.

> What I actually had in mind, though, was more like a modified ldconfig
> that would create the necessary links in /usr/lib instead of
> /mnt/*/usr/lib "when necessary" (either after failed link creation, or
> by working out the fstype directly).  Implementation is obviously an
> issue, but would the end result be something desirable, or is the idea
> itself wrong?

In my opinion thats a gross hack.  What if I dont *want* it fucking with
/usr/lib?  What if I want to just run the binaries there on the card
with a proper LD_LIBRARY_PATH without root linking?  You're making
assumptions about the users environment.  I'm not going to sacrifice
flexibility or power to work around something.  Thats bad policy.

> > Now, if ipkg did the linking to root it could possibly do that, but
> > there would still be symlinks missing from the card that should have
> > been there, which *should* cause a failure.
> 
> I don't understand.  The loader will fail in the case where
> LD_LIBRARY_PATH points to /mnt/card/usr/lib?  Or do you mean that there
> ought to be some obvious, user-facing error when ipkg can't install a
> symlink to a VFAT filesystem?

Right.  I could see it having valid reasons to reducing the # of
indirections in the symlink paths from the actual shared lib on the card
to the rootfs, but links are still missing that should have existed.
--
Chris Larson - kergoth at handhelds dot org
Embedded Linux Developer - clarson at ti dot com
OpenZaurus Project Maintainer - http://openzaurus.org/


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
Openzaurus-users mailing list
Openzaurus-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openzaurus-users
[prev in list] [next in list] [prev in thread] [next in thread]