Looks like the "X 10" in OS X 10_8_4 is being normalized into "X10" and getting a false hit on a device. So this is one of the drawbacks of stripping out regex/whitespace/symbols. Its not so much the whitespace and symbols which poses a problem, its the regex. Standard data structures cannot index or key on a regex expression. So we would have to resort to something a bit more exotic like what dclass does or we would have to do a brute force search. Im leaning towards eliminating regex from the patterns and re introducing whitespace and symbols into the pattern matching (this will greatly increase the accuracy and eliminate things like "X 10" => "X10").
Does this make sense? Thoughts? ________________________________ From: Radu Cotescu <[email protected]> To: devicemap-dev <[email protected]> Sent: Monday, June 24, 2013 4:26 PM Subject: devicemap java client classify error Hi, Just out of curiosity I've tested the UA Chrome sends on my MBP (Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/27.0.1453.116 Safari/537.36) using the devicemap-java client. Unfortunately it was identified as a Sony Xperia Phone. However the DeviceMap service at http://devicemap-vm.apache.org/javaservice.html identifies my UA as a desktop device. I've tried switching the data files but to no avail. There must be a parsing error somewhere. Regards, Radu
