Hi Michael,

On Wednesday 24 October 2007, Michael Biebl wrote:
> Its hard to estimate such numbers. But given, that today more than 50%
> of PCs are laptops, and with dell having a market share around 20%, this
> would make 10% of users with a dell laptop.

That still leaves 90% of systems where the package is installed for no good 
reason (and 100% of desktops). And I would be very surprised if _all_ Dell 
laptops actually need this.
I agree that Dell laptops are not uncommon, but I still feel strongly that 
this is insufficient justification to bloat every other system.

On Wednesday 24 October 2007, Michael Biebl wrote:
> We were discussing this issue among the hal maintainers before the
> upload. Fact is, that the Suggests mechanism isn't really nicely
> integrated into the apt frontend, so basically noone (at least not
> novices) make use of that. They mostly only notice, when things don't
> work as expected. libsmbios-bin is a rather small package which has
> reasonable dependencies (libstdc++, libxml2), so our reasoning was, that
> in a matter of "just works", it would make sense to make it a
> recommends.

All that is still no reason to ignore the definition of recommends:
   This declares a strong, but not absolute, dependency. 
   The Recommends field should list packages that would be found together
   with this one in **all but unusual** installations. 

IMO this is just not true in this case.

> Advanced users can easily chose to not install libsmbios-bin. 

The only problem is that they won't know _why_ libsmbios-bin is being pulled 
in, except by digging through the changelog.

> Ideally, we should have some kind of support for "hardware based
> dependencies/recommendations". Unfortunately right now we don't have
> something like that.

Petter Reinholdson is working on that. See the _new_ functionality of 
discover. This kind of thing fits _exactly_ in what discover is supposed to 
do now.

Cheers,
FJP

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to