From: Thomas Martitz <thomas.martitz_at_student.htw-berlin.de>
Date: Wed, 27 Oct 2010 01:04:00 +0200

On 27.10.2010 00:45, Paul Louden wrote:

> DPI mismatch is only relevant at all if we're going to be creating
> different versions of the screen for each DPI of device at each
> resolution. If the screen is only to have one version per resolution,
> it's obvious that the DPI excuse is irrelevant since DPI is already
> ruled out as a necessary concern, or we'd be intent on having
> different versions for different DPI.

That's not true. Not making different versions for different DPI doesn't
mean DPI differences can be ignored completely. E.g. buttons need to be
big enough for the highest DPI screen so you can push it with the thumb,
but that buttons will look ugly and huge on a low DPI screen and super
easy to click on with the mouse. Thus you cannot really measure whether
buttons are big enough without actual access to the device's screen.

You seem mixing up "Taking DPI into account for multiple versions of the
design" with "Taking DPI differences between the design target and
design test machine". That's not the same.

I've done work on touchscreen adoption of various parts and touchscreen
cabbie ports so I can tell you that you need to consider this, and that
testing the design on the actual device always reveals some flaws that
weren't visible on the sim.

That said, ideally we should also take DPI into account in the code and
themes (so multiple versions for the same resolution), but I'm afraid
that's not feasible because we have too little developers interested in
GUI work.

Best regards.
Received on 2010-10-27