HAL and scanners

[abel deuring]

Richard Hughes wrote:
> On Fri, 2006-12-29 at 10:30 -0500, Joe Shaw wrote:
>>>I don't think so.
>>I disagree here.  We should always collect and present as much 
>>information about a device as possible, even if they're not yet 
>>supported by software.  The only thing more aggravating than attaching
>>a 
>>piece of hardware and having it not work is attaching a piece of 
>>hardware that is identified as "Unknown device" and having it not
>>work.
> 
> Sure I can agree with this to some level - but i really think we should
> only expose keys that convey useful information in this case; for
> example:
> 
> scanner.is_supported = false
> 
> and not have:
> 
> scanner.driver = "unknown"

scanner.is_supported would be a good solution, but: if we allow more
than one API or application in the scanner namespace (examples: Sane
and vuescan; perhaps also Étienne's gnomescan API), the support
information would probably come from different fdi files. Does HAL
provide a sort of an "or" menchanism so that scanner.is_supported
can be set to true, if the device is supported at least by one API,
or should we use scanner.sane.is_supported,
scanner.api-xyz.is_supported etc?

Abel