As you may notice, it's the only case where "OS DataSource" was populated, but compared to Windows PCs most Macs were also properly recognized in the past by OpenDDR/DM 1.0;-)
A bit of "Magic" the simple parser solution so far does not have an equivalent for are all the *Builder implementations. Some contain extra regognition logic in cases where the UA may be to vague, ambuguious, etc. I am not 100% if the "OS" file is really used, but the special IOSDeviceBuilder (*DeviceBuilder are related to the "Device" file) or similar ones for other devices improve recognition in the W3C implementation. Werner On Wed, Sep 10, 2014 at 3:31 PM, Reza <[email protected]> wrote: > Yes, that is exactly my concern regarding changing the id of > desktopDevice. Also, dClass uses this id in its default configuration, so > it would break downstream data consumers too...! > > There is no performance penalty for this. The aliases are resolved once > and compiled away when the pattern index is being created. > > "genericDesktop" was simply created so all of the different desktops could > have a common parent for shared attributes. It was never meant to be used > as a detectable device. But the plan is to eventually rename desktopDevice > to genericDesktop since the latter is a better naming convention. > > > ________________________________ > From: Werner Keil <[email protected]> > To: "[email protected]" < > [email protected]> > Sent: Wednesday, September 10, 2014 8:59 AM > Subject: Re: Data > > > Well I just ran a quick search for the keywords, and especially test data > of all kinds (in client modules or a separate artifact) contains > "desktopDevice" about 10.000 times! > No more than 10 references of "genericDesktop", so I must admit, Eberhard > has a point this aliasing probably should have been witheld till a 1.5 or > 2.x release;-| > > If nobody comes across a sustainable performance penalty by the alias > compared to say: > <device id="windowsDesktop" parentId="desktopDevice"> > > we can probably live with it for now. > > Werner > > > > > On Wed, Sep 10, 2014 at 2:36 PM, Reza Naghibi < > [email protected]> wrote: > > > It's probably best to wait for 2.0 to change desktopDevice since it will > > be a a rather large breaking change. > > > > > > > > <div>-------- Original message --------</div><div>From: Werner Keil < > > [email protected]> </div><div>Date:09/10/2014 8:05 AM (GMT-05:00) > > </div><div>To: [email protected] </div><div>Subject: > Re: > > Data </div><div> > > </div>I did not suggest moving the "root" of desktop, only tha alias at > > most. > > Unless there was a reason for keeping the alias in the patch file (like > > some clients in need of it?) I'd say we remove it no later than 1.0.2 > and > > stick to "genericDesktop". WDYT? > > > > Almost every item in the main data file is "*generic", and as you say > > people can still add one with their own name in the patch file. Even the > > "patchDevice" which is an empty shell seems no more than a stub. If we > keep > > that, I suggest we put a little bit of XML documentation there to explain > > how people can extend device signatures there. > > > > For e.g. the 600x800 screen a custom extension makes sense, but there's > no > > reason not to keep a default signature in the other file. > > We don't need to follow WURFL's path where we can do better, the > > performance of the W3C implementation already showed ODDR/DeviceMap does > a > > lot better there;-) > > > > Werner > > > > On Wed, Sep 10, 2014 at 1:57 PM, Reza Naghibi < > > [email protected]> wrote: > > > > > Remember, any core device or pattern can be changed by defining the Id > in > > > the patch file and then overriding the value. > > > > > > I don't think that any of our definitions should live in the patchfile. > > > Wurfl kept desktops in the patchfile because it did a poor job of > > detecting > > > them. That's not the case for us which is why we expanded the core > > > definitions to bots and other device classes. > > > > > > Patching should only be required to add functionality. If we are > missing > > > functionality... we should consider fixing it in the core definitions. > > > > > > So I'm for keeping all device and pattern definitions in the core > files. > > > > > > > > > > > > <div>-------- Original message --------</div><div>From: Werner Keil < > > > [email protected]> </div><div>Date:09/10/2014 6:27 AM > (GMT-05:00) > > > </div><div>To: [email protected] </div><div>Subject: > > Re: > > > Data </div><div> > > > </div>It would be great to allow adjustment in the "Patch" file by > > moving a > > > device definition like "desktopDevice" there. > > > Whether it's named "desktopDevice" or "patchDesktop..." or similar does > > not > > > really matter. > > > > > > It seems not show-stopper for 1.0.1, as long as the W3C compliant > builder > > > signatures work now, I would not insist on changing it now, but we may > > want > > > to clean that up in the next release. > > > > > > Werner > > > > > > On Wed, Sep 10, 2014 at 1:35 AM, eberhard speer jr. <[email protected]> > > > wrote: > > > > > > > -----BEGIN PGP SIGNED MESSAGE----- > > > > Hash: SHA1 > > > > > > > > RIM - BlackBird...got it, no change required. Easy. > > > > > > > > And iOS and iPhone OS ? similar case ? > > > > > > > > esjr > > > > -----BEGIN PGP SIGNATURE----- > > > > Version: GnuPG v2.0.22 (MingW32) > > > > > > > > iQEcBAEBAgAGBQJUD465AAoJEOxywXcFLKYcgFEH/0SFzkl3h8WefhOZWM7tIAIj > > > > SawluAF1iBugkaP+G62zafK2h7OS0ihjRnic/+lcvxIxzlUkDQnt1yy2OZqDeIhm > > > > jS7tM0VAfkykHcvgB1yolRxUSRtAT/rGdAeUnIzojqqR7C4+FUXE2anVvwFEcGqz > > > > 27UAX/vRkq3Pq89ZBLcu7CTMJEsbzMFxQvL9Uelf4JWyWSeKemlV2Bc/KKa8j9Uc > > > > s2ucwS2KFY5tFYK1QllhUEhQEOgVNKxi8K8DLoc+KI00b5w3HMBif46ZJ/Yw4vpC > > > > b1duhGZToVcbku7EgbhcjjDYxtSiO5ROzhm6iT2zvgWo93ZXWm0yHpbDYhZs47E= > > > > =d7Ls > > > > -----END PGP SIGNATURE----- > > > > > > > > > >
