polling vs initial check of a device

[Alexey Morozov]

David Zeuthen writes:

>> This would allow to from one hand remove polling from 'proper' USB 
>> flash drives, which don't require it.
>>
>> After looking at the hald code (as of 0.4.7), 
>
>
> You should rather be looking at HEAD (which has just grown a D-BUS 
> 0.30 dependency, sorry).

Ouch... Ok ;-)... If I remember correctly, dbus-0.30 has certain API 
changes so end-user apps such as gnome-volume-manager or KDE volume 
manager will break w/ it, right? (Yes, I know that the latter is 
orphaned in favour of new media:/ handler in upcoming KDE release, but 
it nevertheless works for me in KDE-3.3 environment).

>> It seems that the kernel manager of my distro has already fixed the 
>> ide-cs issue (and now he's in process of publishing his patch :-)),
>
>
> Note that hal HEAD handles this nicely as we simply ignore hotplug 
> rem/add from IDE devices. Of course, this is a hack, with that patch 
> we would not need that.

Ok. Hopefully the patch author will officially publish it within few 
days so it will go to the vanilla kernel tree

> I think what we need are two options
>
>  storage.media_detection_enabled  <-- whether we should poll
>  storage.check_for_filesystem    <-- whether we should check for fs
>
> For ide-cs we would set the m_d_e=FALSE and c_f_f=TRUE. For x86 legacy 
> floppies, LS120 drives and IDE zip drives we would have m_d_e=FALSE 
> and c_f_f=FALSE (since they are just floppy drives in effect).
>
> How about that?

I don't mind. The main question is who would implement that? ;-). It 
seems that I could do that but if HEAD changed drastically from 0.4.7, 
it may took some time for me to recognize proper places to apply these 
changes.

Alexey Morozov
_______________________________________________
hal mailing list
hal at lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/hal