Re: Screensaver support for the Hurd console

[Marco Gerards]

Marcus Brinkmann <marcus.brinkmann@ruhr-uni-bochum.de> writes:

> At Sat, 08 Jan 2005 20:16:23 +0000,
> Marco Gerards wrote:
>> All work has mostly be done.  At the moment my biggest doubt is how to
>> configure the screensaver support when starting the console client.
>
> First of all, do we want to allow multiple screen savers to be present
> at the same time, and if yes, what is that supposed to mean?

As I proposed it they will be run in sequence.

> Having graphical eye candy kick in after a minute, and DPMS after two
> minutes, sounds genuinely useful to me, so limiting to only one
> screensaver driver seems to be not flexible enough.

Right, that is what I proposed.  Before switching the monitor off it
is nice to see some eye candy.  Switching off the monitor, etc really
early is not sane because you have to consider monitor startup times.

> Also, if you have multiple hardware displays, each display may need
> its own screensaver driver.

Do we really want that?  It sounds complex to me, in that case we need
a really complex way to configure the screensavers.  I don't think it
is sane to use a single console client for multiple displays anyway,
but perhaps this is short sighted.  Can you tell me more about this
and especially how you want this to work.

> However, if you have an eye candy screensaver, surely only one can run
> at a time.  Should we enforce this, and if yes, how?

What I do in my `life' screensaver is iterating over all display
drivers to write the display too.  In that case, even when you have
eye candy screensavers, it works with all hardware.  Just the specific
screensavers only work on one type of device.

And even in the case of two vga cards, the same screensaver is used
and because of that both screens will be blanked (when all of this is
implemented), right?  So the only thing that currently won't work is
vga + ncursesw + DPMS VGA blanking.

> Now, about the configuration of timeout values.  This actually depends
> to a substantial extend on the implementation, and I don't remember
> the details anymore.  So, I actually would like to review the
> implementation first.

I described how it worked, why is the implementation important?

Thanks,
Marco