[colord] Are display device IDs still unique if EDID contains no 'ASCII' serial?

Florian Höch lists+colord at hoech.org
Mon Dec 19 12:04:28 PST 2011


Am 17.12.2011 00:58, schrieb Richard Hughes:
> https://gitorious.org/colord/master/blobs/master/doc/device-and-profile-naming-spec.txt
> for all the juicy details.

Yes, that's what I'm using :)

>> I looked at the raw EDID of my screens, all from different manufacturers,
>> and they all have the same serial at bytes 12...15 (16843009, hex 01010101).
>> Most of them except my Philips LCD TV have a unique non-generic monitor name
>> and ASCII serial though.
>
> Some EDID data really sucks.

Indeed. Though I think now it will only be a problem in corner cases.

>> I'm wondering if, in the case that there is no ASCII serial available from
>> EDID and the monitor name is very generic, could it be possible to introduce
>> duplicate device ID entries in colord's database if several of these devices
>> are connected?
>
> Well, if you try to create a duplicate device, you get an error. If
> there are two monitors attached with the same serial number then we
> really just use the same profile for both of them, as to colord they
> are the *same* device.
>
>> 'xrandr_Philips_Consumer_Electronics_Company_Philips_FTV_16843009', so
>> basically I'm wondering what's going to happen if I e.g. had another Philips
>> TV connected (but I only have the one, so I can't try it).
>> Has this been thought of?
>
> Yes and no. There's really not a lot we can do, other than blacklist
> EDIDs and fall back to the xrandr-{%xrandr_name} format.

Agreed, not much that can be done. I think problematic EDIDs may mostly 
be LCD-TVs (my previous Samsung TV also had a generic 'Samsung LCD TV' 
monitor name and no usable serial). The (few) desktop screen EDIDs I 
looked at had reasonable monitor names so far.

Florian


More information about the colord mailing list