KStandardDirs and Autopackage

Andy Fawcett andy at athame.co.uk
Wed Apr 14 17:45:22 BST 2004


On Wednesday 14 April 2004 00:31, Arend van Beelen jr. wrote:
> Hi!
>
> Maybe you've heard of the tool autopackage (http://autopackage.org).
> It's a tool that allows for distribution-neutral binary packaging.
> Currently we're readying Apollon for shipping it as an autopackage.
> Because of it's goal of cross-distro binaries it does require some
> changes to application though, basically to avoid hard-coded paths in
> the binaries. Fortunately KDE applications already use little
> hard-coded paths because they use KStandardDirs for locating files.
> Therefore, it's quite trivial (once you know it) to integrate
> KStandardDirs with autopackage so that KSD will not only scan KDE
> dirs, but also the directory into which the application was installed
> (not necessarily the main KDE dir anymore). Of course, the best
> solution would be to include the small library used (BinReloc) with
> KStandardDirs, so that application developers will automatically and
> transparently use this library. It won't require any changes to the
> API of KStandardDirs, it will just add a default location to search
> (including the logic that determines what that default is).
> Now, do you think it would be a good idea to add this to
> KStandardDirs or would you prefer a different solution? FYI, A
> similar proposal has been sent to the GLib developers (though it's a
> bit more tough for them as they don't have a KStandardDirs-like API
> in place at all)

From the status section of the homepage:

"autopackage is currently not API stable. Packages built with one 
version will not work with newer nor older versions. Until we reach 
greater interface stability it is not recommended that you use 
autopackage for your own software."

Now, ignoring the above for a moment, your idea is basically sound. I've 
not looked much further than the homepage at autopackage.org, but I 
couldn't see anything about portability to non-linux systems (it may be 
that there's nothing stopping it being portable, I won't know until I 
dig a little deeper)

*When* the API is stable, and providing it can be integrated without 
detrimental effect on non-linux users, I think it's a reasonable idea. 

Then again, I'm (just) one of the KDE-FreeBSD packagers, so binary 
portability isn't the be-all and end-all for us anyway :)

A.

-- 
Andy Fawcett                                     | andy at athame.co.uk
                                                 | tap at kde.org
"In an open world without walls and fences,      | tap at lspace.org
  we wouldn't need Windows and Gates."  -- anon  | tap at fruitsalad.org




More information about the kde-core-devel mailing list