Blackberry Webkit family not being detected as mobile about ua-parser

[tobie]

Not sure how to best solve this without introducing the idea of a model / version number for devices, eg:

{'is_spider': False, 'is_mobile': True, 'family': 'BlackBerry', model: '9320' }

@elsigh ideas welcomed.

[tobie]

Related to #65.

[dmolsen]

FWIW, the PHP lib classifies this UA as mobile because $ua->family matches a string in my mobile browser array. This info is already available in regexes.yaml as mobile_user_agent_families.

[tobie]

Right, so I was looking for string to be equal, not using stristr.

Doesn't that make the Playbook a mobile device too and break the current test suite?

[tobie]

The distinctions we're making between isMobile and isMobileDevice worries me, actually. It's awfully confusing.

For example, how do you categorize and iPod touch? is it a tablet or a mobile device?

What about an iPad mini with 3G and Skype? is it a mobile device or a tablet?

What's next? Whether the device is touch enabled or not?

Screen size?

Has a camera?

Etc.

When do we cross the line and fall into territory that belongs in a project like WURFL or the like? (Note that I think a good, open-source version of such a project would be awesome, it feels like its out of scope).

/cc @elsigh

[dmolsen]

I think it's sort of confusing as well. There's a reason I haven't pushed the isMobile etc. back upstream myself. They seemed to be a departure from the original intent of ua-parser which is why I called my lib a pseudo-port (and it still is).

• isMobile should be thought of as a mobile browser not on a tablet.
• isMobileDevice can be either a phone or tablet
• isTablet is just terribly brittle. The really bad example is Windows Tablet which can return true for devices that, if we were just looking at them, we'd classify as netbooks. They just happen to have the right OS/under-the-hood-hardware combo to affect the UA. That and the screen dimensions open a whole other can of worms.

As for the examples cited:

• Playbook is true for isMobileDevice and isTablet and false for isMobile so doesn't break the classification.
• iPad (mini or not since the UA doesn't provide a distinction) is true for isMobileDevice and isTablet
• iPod touch is true for isMobileDevice and isMobile

Frankly, I was thinking of dropping isTablet. I think it gives false hope. isMobileDevice could then also be dropped and isMobile then simply refers to what I classify as mobile browsers. It can be expanded to be set to true for devices we know are mobile but can't assign a browser family too. The problem there though is that, in my set-up, isMobile would be true for the Playbook because I list BlackBerry WebKit as a mobile browser which the Playbook is obviously using. To me that would be an appropriate classification. You may disagree :)

FWIW, I haven't properly integrated the test suite. Doing that during the holiday weekend.

// just because i can include it... 🎱

[tobie]

This is a massive can of worms. Think we should just stay out of it.

On inspection of the python code source I notice @elsigh is also doing strict equality comparisons with the mobile browser array.

[elsigh]

Oops, I shouldn't have pulled that ;0
Trigger happy holidays.

Yeah, ultimately, I agree that these distinctions are troublesome.
What does "mobile" even mean anymore?

I'd be ok with just reverting this concept from ua-parser.

[tobie]

So what should we do: just have a device object with a single, family property?

[tobie]

Closing. Decided to remove all boolean props from the JS version.