uswsusp or not?

[James Westby]

On Thu, 2008-09-25 at 11:56 -0700, Steve Langasek wrote:
> On Thu, Sep 25, 2008 at 02:27:01PM +0100, James Westby wrote:
> 
> >   * When recommends became installed by default the recommends
> >     of uswsusp by pm-utils meant that everyone got uswsusp
> >     installed.
> 
> This definitely didn't happen for me (or, as far as I can tell from bug
> report frequency, the vast majority of users?)

Yeah, I'm not sure what it is that causes it to affect someone yet.

> > should we create a dummy s2ram that uses the kernel method instead?
> 
> That seems like a reasonable option if it's practical.  Alternatively, would
> it be feasible to change pm-utils to choose its backend based on which
> method (suspend vs. hibernate) was selected?

Yes, it would be feasible, but it would be quite an intrusive change at
this stage of the release cycle. It does seem sensible to me anyway,
but there may be something I'm not missing.

> > If we don't want uswsusp to be used by default then the recommends
> > should be downgraded to a suggests
> 
> I agree with this, in any case.

I will prepare an upload for this.

Does anyone know if this will remove the package for those who have
the package installed automatically as a Recommends of pm-utils?

> > A bit more investigation shows that this doesn't affect new
> > installs, as uswsusp is not on the CD. My current guess is
> > this only affects machines that were installed when pm-utils
> > wasn't installed by default and upgraded to newer releases.
> 
> I'm not sure it affects anyone who didn't install the uswsusp package
> manually - pm-utils in hardy already had the Recommends on uswsusp, and hal
> in hardy Depends: pm-utils.

% aptitude why uswsusp
i   hal      Depends    pm-utils
i A pm-utils Recommends uswsusp 

which is the output I have been basing my investigations on. I appear
to only have uswsusp installed because of the Recommends, and it
is marked automatically installed, so I didn't explicitly choose to
install it. I don't know "aptitude why" well enough to completely
trust this interpretation of its output though.

Thanks,

James