extension of lsb packages

[Jason Gunthorpe]

On Mon, 4 Mar 2002, Jim Knoble wrote:

> Don't be pedantic.  The leading 'lsb-' is because someone thought it
> would be a good idea to put 'lsb-' in front of every LSB-compliant
> package.  It's not.  It's too long, and it's got too many lsb's, and

Yes, and I have long advocated doing that. It is important that the entire
LSB mess be isolated from the OS vendor supplied packages and from other
LSB vendor packages. Since the choice of implementation must work with
exsting technology (ie RPM) the only choice is to prefix the package name
with 'lsb-' and perhaps the vendor name too. Once you've done that you
might as well carry that into the filename too.

> Many package names are already long enough as it is.  Making them longer:
> 
>   lsb-apple-quicktime-core-13.2-1.update2.i386.lsb
> 
> (simply by way of hypothetical example) is a bit too much, don't you
> think?

I have advocated doing this as well. I belive that it is very likely that
multiple vendors will be providing competing LSB compliant components for
some time. It is essential that this be allowed to happen and the only way
it can is if we encode this information in the package name so
dependencies can work properly. 

Your case is misleading because Apple is likely to be the only vendor of a
quicktime package. Consider something like gnome or kde. We have OS vendor
gnome (RH, Debian, etc), Ximian gnome and Sun gnome.  All must be able to
exist at once on the same system, so they all need different names.

> : We could possibly have non-lsb packages using the lsb format,
> : particularly if the lsb format becomes something more than it is..
> 
> Who cares?  The package manager simply cares whether the file format is
> correct, and whether the dependencies are fulfilled.  If the

It also has to be able to distinguish RedHat, LSB and non-LSB versions of
the same package. Thats what the lsb- is for.

Jason