Re: Mapping browsers to 2.2.1 Dictionary Browser Members

Niels Leenheer schreef op 2013-10-18 10:58:
> On Oct 17, 2013, at 10:19 PM, Ronald Mansveld 
> <ronald@ronaldmansveld.nl> wrote:
> 
>> Hi all,
>> 
>> As Doug mentioned in the earlier email, I'm currently writing the 
>> spec for the datamodel to exchange compatibility data.
>> 
>> The first draft can be found on: 
>> http://webplatform.ronaldmansveld.nl/compat_tables_datamodel.html [1].
> 
> I'm thinking the definition of a browser in 2.2.1 may be too
> simplistic. Not every test result I have may map properly to this
> definition.
> 
> First of all, a couple of small nitpicks:
> - I think you should separate vendor into vendor and name. For
> example "Google" is a vendor, while "Chrome" is the name.
> - The example for Chrome you have includes the full version, while IE
> includes only the major version. If would suggest to use the least
> specific version possible that do not include bug fix releases. So IE
> would be 8, 9 or 10. Chrome would be 25, 26, 27, etc. And Opera 12.00,
> 12.10, 15, 16. Firefox 3.5, 3.6, 4, 5, 6, etc.
> 
> MAPPING TO OPERATING SYSTEMSSecondly, I do not think a browser map to
> a single operating system. You can have combination that are less
> specific and that are more specific, for example:
> 
> Chrome 30 on Windows and OS X can be considered the same.
> Chrome 30 on Android is separate.
> Chrome 30 on iOS is separate.
> => We need to be able to specify multiple OSes for a single browser
> 
> Internet Explorer 11 on Windows 7 or Windows 8 are definitely not the 
> same.
> => Just "Windows" is not enough, we need to be able specify a version
> 
> BROWSERS WITHOUT NAMES AND NICKNAMES
> Not all browsers have an official name. Or sometimes it does, but we
> use the name of the OS instead. For example:
> 
> Safari for iOS is usually called just iOS with the version number of 
> the OS.
> The stock browser on Android does not have a name and is usually
> called "Android" with the version number of the OS.
> The Playstation 3 also has a browser, but it does not have a name. It
> is usually just called 'Playstation 3': the name of the device.
> 
> BROWSERS WITHOUT OS, BUT WITH FIRMWARE
> Not all browsers have an OS. Okay, technically every application runs
> on an OS of some sort, but in case of devices, the OS is not always
> something that can be separately named. For example: the Nintendo Wii
> just runs a specific version of firmware, not an OS that can be
> separated from the hardware.

Referring to the earlier conversation with Tobie: I'm lenient to using 
the UA-string, if the license of the UA-parser is compatible. That way 
the UA-string is the only thing needed, and all needed data like 
versions, names, OS'es etc. can be parsed from there. As far as I'm 
concerned that would solve all the points above.

Received on Friday, 18 October 2013 10:53:29 UTC