Chen, Congwu wrote:
Jussi Kukkonen wrote:
> Chen, Congwu wrote:
>> Ohly, Patrick wrote:
>>> On Tue, 2010-03-23 at 01:30 +0000, Chen, Congwu wrote:
>>>> Ohly, Patrick wrote:
>>>>> For example, suppose we have "fingerPrint = Nokia 7120c, Nokia
N85".
>>>>> I would find it confusing to have a "Nokia N85" and then
being told by
>>>>> the GUI tells me that "we think you have a Nokia 7120c".
>>>> I think the GUI can still tell you "we think you have a Nokia N85,
because
> it
>>>> is a perfect match and we know the device name is Nokia N85, though the
>>>> template name is Nokia 7210c.
>>> Does the GUI get enough information for this already? If yes, can you
>>> describe how Jussi can implement this? If no, what needs to be changed
>>> by whom?
>> Yes, I think so. The temporary template has 'deviceName' and
'fingerPrint'
> property
>> for this two.
> Congwu, the problem is this: my phones deviceName is 'jku'. If the UI
> does what you suggest we end up with:
>
> "We think your device is a 'jku'. If this is not correct, please
> take a look at the list and select your device if you see it."
No, what I suggested is only for a perfect match, as the example
above. For your case, the match score will be 0 or a relatively
small number, what you might present maybe:
"We don't know your device or we think your device looks like.."
I wasn't talking about the devices we currently support, I'm talking
about the general case. There's very little point in coding UIs for just
"Nokia 7210c" if it's not usable for all S40 phones we will support.
I agree if we have can automatically detect 'model name' for
a device, we can provide much better matching capability, but this
is not supported currently and need more work to do: using
default device name, bluetooth profile, SymcML DevInf....
In your case, if we can detect 'jku' has a model name 'Nokia 7210c'
and even have the device type 'S40', we can provide a much better
matching result.
This isn't about matching really (as important as it is). This is about
presenting the results when we have a match. A model name would be nice
there.
> I don't think that's acceptable. If we support the phone
model I have,
> it would be nice if the UI told me that, but I won't do it if it lead to
> UIs like that. Don't get me wrong, the current info is quite usable: I'm
> fine with just showing the template name for now if the model name seems
> difficult with the schedule.
Yes, detection the model name for a phone would need additional work and
I don't think it can get into SyncEvolution 1.0
Ok.
> That said, here's my thoughts on this -- something I should
probably
> have written a while ago. I think this would go a long way: All
> templates would have:
> - template name (always)
> - user identifiable device name (always, this is currently
> "deviceName")
> - score (always, this changes depending on the device this
> template is for)
> - list of all model names for this template (always)
> - current model name (when available, syncevolution sets this when it
> has an idea what the model is, either from BT-name or from other
> future datasources, like the Bluetooth profile we discussed)
I think the only missing part is 'current model name'. Get the information
from BT-name seems problematic, how do you know this is a model name
not a user defined name? From Bluetooth profile can help detect the model
name if the phone supports and from DevInf we may even detect the type
of the phone.
Well, I imagine it would go like this:
* Get device specific data that might help:
- the BT profile
- DevInf
- bt name
- other?
* go through all templates, look for matches between the device
specific data and the "supported models" list in the template
* if you get a good enough hit return the "supported model" as the model
name
So user defined name would be used to match but would never be returned
as a model name.
Jussi