We can't say what they do now or other (W3C compliant but closed-source) competitors like MaDDR or DeviceAtlas, but the WURFL snapshot Bertrand conserved shows just a few slightly more cryptic IDs like "sie_s45_ver1_sub00_102" while the vast majority of IDs are things like "j2me" or "streaming", hence very human-readable, too[?]
Cleaning up stuff that was maybe manually added or derived from similar sources (OMA, that's where WURFL got its seed prior to closing it down) is OK. If entries were to be maintained in a sort of DB-structure and a service or "portal" allows brand new entries, then some sort of "surrogate" could be OK, if it's automatically generated by the DBMS or server. I don't think we must have more than one ID, if someting less human-readable (like that "sie_s45_ver1_sub00_102" which was also seemingly generated) is added that's OK, but maintaining 2 separate IDs I am not sure about that[?] Werner On Mon, Aug 25, 2014 at 9:14 PM, eberhard speer jr. <[email protected]> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > So, no one sees the sanity of having an identifier that 'travels' well > [think http, think XML, think json...encoding] besides a > human-readable pseudo-identifier [common practice] ? > Despite the fact that the DeviceMap [OpenDDR] data itself already > clearly shows signs of having bitten by that 'bug' [barnes & noble vs. > barnes and noble] in *non*-key data... > > I am baffled. > > esjr > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v2.0.22 (MingW32) > > iQEcBAEBAgAGBQJT+4skAAoJEOxywXcFLKYc3ykH/0oXmlmrvh0nTGUgjUH8R9nh > UHGVsuHDiU0DG2J9aoFW0a3WvJUKfeQWuKO4zaHzkegD1fbtBgOTYTX//kNa74RJ > yMhyU2yc75ZnPUjUd7t3/wZwW1as74wTgDftOA72j+EyXR1dyZwssEw1K4kRnXI8 > F/QJqe4QaR7dxvp5x2SpkuDJ4R9Ypohz5dNcNuI4ImkGtBQYlIe0K0JIMbBvpWq4 > 7dxYs+dBzM5F2ZM4hA2oGTGTg+/JsNP0vB8TWd5gxTwFzvz7kE3jip49qiRRVJlw > ZSYW6KqmXEiqMNgmWkHvxo0BZceknwNZjrKtgdIOAoea0/2LJ4+eKcKPZn9vM7g= > =muRs > -----END PGP SIGNATURE----- >
