Re: Screen scaling (was A bug?)

[Russell Shaw <rjshaw@iprimus.com.au>]

Daniel B. wrote:
> Russell Shaw wrote:
>
> > ...
> > Doesn't that mean that any assumptions about font and window sizes by
> > X apps is all wrong? Instead of 1pt being 1/72", it's more like 1/60",
>
>
> Actually, 1 point is always 1/72".  (That's the definition of a point.)

That's what i mean. If you measure 1pt with a ruler against the screen,
you should see 1/72" regardless of whether the monitor is set to 640x480
or 1024x768. Resolution should only have the effect of making diagonal
lines and curves less smoother. With 100dpi on my system, 1/72" ends up
being physically 1/60".

> But you are absolutely correct that many assumptions about displayed 
> font sizes are wrong.
>
> > ... It also means that
> > changing the resolution from 800x600 to 1024x768 changes the *physical*
> > screen size, 
>
> Do you mean the screen size of a font (the size of a font on the screen),
> or the screen size (the size of the screen)?

It will cause the physical size (what you measure with a ruler) of fonts
and windows to change. The physical dimensions of your monitor as reported
by xdpyinfo will also be wrong.

> > making all the fonts and everything else change size when
> > they should be constant.
>
> Which sizes do you think should be constant?
>
> If I increase my screen resolution (number of pixels), it's usually
> because I want more information on my screen, so (usually) everything
> should stay the same size in pixels but shrink in physical size on the
> screen.

No. That's broken microsoft behaviour. Increasing the resolution should
only make angled lines and curves smoother. If you want everything on
the desktop like menus and icons to be a different size, there should
be a setting in the window manager, or you could use startx -- -dpi 100
to change everything including the font sizes.

> (Of course, there are other cases too.  One might increase screen
> resolution to display something more smoothly but still want physical
> sizes of everything to stay the same.)
>
> I think the problem is that we don't distinguish between the specified
> printing size of text and the currently displayed size of text.
> (Actually, that second part is "the size on the display of the screen
> font currently used to display text with the specified printing size.")

By specifying DisplaySize in XF86Config-4, fonts should all *look* the
same size on different sized monitors. However, because gui toolkits
specify most things in pixels, then windows and menus should change
size with the size of the monitor.

> To me, a "10-point font" is one that prints on paper with a size of 10
> points (if no further scaling is done when printing).  Obviously, text
> set to print at 10 points can appear at any size on the screen.
>
> It makes sense to talk about a "10-point screen font" only when it's
> calibrated for the screen resolution in pixels and for the physical
> screen size.

I read somewhere (when doing M$ programming) that the physical size
displayed for a font on the screen is bigger than the same font on
a printer, to compensate that monitors are usually read further away.
This would be acceptable.

> Regarding assumptions' being wrong, one major peeve is MS Outlook and
> its users.  When the default "12-point" font size looks too big on their
> screens, they think that it's appropriate to set the font size to "10
> points."  That causes Outlook to send HTML mail with an HTML font size of
> 2.  My browser/mailer is adjusted to display normal-sized text (size=3)
> at a comfortable screen size.  Of course, when I get the Outlook-
> originated mail with a reduced HTML font size, it displays uncomfortably
> small on my screen.  ...all because Bill Gates and company don't provide
> a way to display message text in a smaller screen font on the local 
> machine without screwing with the HTML font size.

I see some mail like that in mozilla, but i have it set so that
when i hit reply, the message is converted to plain ascii which
is easier to read.