I demand that Ville Skyttä may or may not have written...
On Sunday 17 February 2008, Matthias Schwarzott wrote:
On Sonntag, 17. Februar 2008, Joerg Bornkessel wrote:
We detect this in the vdr-xineliboutput-*.ebuild an let install the xineplug_inp_*.so to the given dir by the headerfile. There is nothing wrong.
Yes there is – your detection method.
Jussy, just skip the xine-lib-1.1.10.1 Version, use an older version, or i hope it will fixed by xine team, then the next version.
Bad advice, unless the known potential for remote exploits (admittedly remote, but let's ignore that) isn't a problem for that particular installation.
This is fixed in ebuild vdr-xineliboutput-1.0.0_rc2_p20080120-r1. Now we restrict the xine-lib version to 3 components.
That'll be broken again by the next release (I've changed the naming), but there'll be a benefit: once the plugin is rebuilt, you'll be able to avoid rebuilding it unless you have need of some new ABI extension.
Or at least that's the plan :-)
(Incidentally, it was unrebuilt vdr-plugin-xineliboutput blocking xine-lib migration into Debian testing – needed for security fixes – which prompted this change...)
FWIW, I'm pretty certain that the only upstream supported way to retrieve the dir where to install xine plugins is to use "xine-config --plugindir" or "pkg-config --variable=plugindir libxine".
They are, but you should not use pkg-config unless your plugin is not 1.1-compatible.