Re: A proposal for units in GTK+

[Keith Packard <keithp keithp com>]

Around 1 o'clock on Dec 2, Michael Meeks wrote:

> So - all this stuff about angular eye slicing, virtual inches,
> foot-pounds etc. is well beyond me; however in OO.o there are some
> people that think when you are at 100% zoom - in the ideal world you
> should be able to hold your printout up to the (flat etc.) screen and
> see the perfect parity between the printed and on-screen page [ if your
> monitor isn't lying about it's DPI ].

That's certainly one view, and used to be true in Macintosh systems.  
However, it's actually not all that useful as the screen is always of 
lower DPI than the printed page, so people tend to want to make text 
larger to make it more legible.  Also, the screen is generally held 
further from the eye than paper, so this again indicates that the default 
text size for computer monitors should be larger than on paper.

> It seems to me with an angular perspective, that instead you have to
> press the sheet of paper to some virtual plane between yourself and the
> display (or beyond) - is that so ?

Precisely; you make the text the same angular size on the screen as it 
would be on a piece of paper held in your hand for reading.

> And/or is it a reasonable thing to want this behavior ? and if so -
> would it be sensible to have a "believe the monitor really" setting in
> the Gnome font settings 'Font rendering details' dialog ?

Yes, we should have a setting where the DPI tracks the monitor precisely 
for times when you really do want to hold the piece of paper up to the 
screen and have them match.  But, that should not be the default.

> Either way - I'm well in favour of the status quo in our OO.o - just
> using the XGetDefault and falling back to the DPI, but others are
> thinking differently; insight appreciated.

I'd rather see us use XGetDefault/XSETTINGS and fall back to DPI computed
from the pixel size of the monitor, let the XSETTINGS value be set to the
physical DPI for cases when people really do want that obvious-but-not-
ideal value.

-keith