Resolution indpendence

Glynn Clements glynn at
Fri Jun 27 13:34:23 PDT 2008

Nicolas Mailhot wrote:

> > The problem as I see it is the conflation of DPI as 'the thing I need to
> > change to make my fonts render at a reasonable size because 12pt is the
> > standard for very reasonably readable text and changing all my documents
> > as stupid' (users care about)
> Actually, no. You have two sources of computer text:
> 1. system applications that should just use the user default font size
> -> have a setting where the user enters his preferred font, the
> preferred  size for it in pt, and have apps obey it (with % sizes for
> titles)
> 2. documents produced by someone else, that may use different prefs
> from the ones of the user
> -> all the apps that manage those documents have a built-in zoom
> system, and it's stupid to even try to correct all those with a
> system-wide dpi kludge because every external document won't use the
> same font sizes anyway and the correction will vary document per
> document (you can try to automate "match to the system font size later
> but it'd be a dynamic document-specific adjustment not a fixed fake
> dpi value)

I still live in hope that people will eventually realise that physical
layout just doesn't make sense when most documents never make it into
hardcopy form.

Really, everything *should* fall under #1. Except that for some of us,
the preferred font size is "as many pixels as are needed in order to
be able to easily distinguish a 'B' from an '8'", and no more.

> > and 'thing which must match my physical
> > properties exactly as I'm doing typesetting' (statistically, no-one
> > cares about this).
> Do you have any study that says users would not like this? They only
> do not care because it's been broken so long (just as they didn't care
> about AA text when the only thing available was pixelated bitmap
> fonts).

I care about AA text. Specifically, it matters quite a lot to me that
I never have to look at it. The only time that scalable fonts make
sense is if you absolutely have to display some text at an exact size
(i.e. desktop publishing). Even in cases where there are *some* layout
constraints (e.g. diagrams), choosing the nearest bitmap font size is
usually perfectly adequate.

Otherwise, I'd really be much happier if it would just use the same
old 12-pixel adobe-helvetica (or adobe-courier for monospace) bitmap
fonts that I've been happy with for the last decade or two.

The notion that AA text is a good idea is a classic example of reverse
reasoning. Starting with the quasi-religious belief that font sizes
should be specified in points, where those "points" reflect the exact
size of the monitor (rather than simply being some kind of abstract
unit), the only place left to go (given that you cannot construct an
infinite number of hand-crafted bitmaps) is that you should use
scalable fonts.

If you start from the fact that hand-crafted bitmaps are more legible
than rasterised vectors (with or without AA; each has different
drawbacks), you might stop to wonder whether using the exact physical
size is *really* worth it. Once you actually consider that issue, it's
pretty obvious that the "physical size is sacrosanct" crowd have it
wrong. Which is why they never consider it.

Glynn Clements <glynn at>

More information about the xorg mailing list