xserver forces 96 DPI on randr-1.2-capable drivers, overriding correct autodetection

Michal Suchanek hramrach at centrum.cz
Sat Sep 24 07:19:29 PDT 2011

On 24 September 2011 12:26, Corbin Simpson <mostawesomedude at gmail.com> wrote:

> Okay, it's time for me to speak up.
> I "sabotaged" this entire effort. You see, five years ago, I had a
> laptop with a brand-spanking-new Radeon X1650. Only fglrx was
> available. fglrx used the actual screen DPI, causing all of my fonts
> to be stupidly large; from login to logout all of my glyphs appeared
> to be anti-aliased Michelin men. When the open drivers started to pick
> up support for my chipset, I couldn't help but notice that all of my
> lettering suddenly reverted to a reasonable size. From then on, I
> vowed that I would never let the bloated, oversized fonts taint my
> screen, nor anybody else's screens, ever again.
> So for the past half-decade I've waged a secret war, carefully
> manipulating key members of the X.org team in order to protect the 96
> DPI "standard" in the name of small fonts everywhere. I feel only
> small amounts of shame at this revelation; I regret nothing.

I don't deny that on your screen or to you personally 96 DPI might
look good. If based on this single configuration you adopt the
somewhat shortsighted position that this same DPI would work as
amazingly for any other screen and make it the default that's fine.
Anyone remembers "640kb should be enough for anybody"?

The bug would not gain nowhere near the momentum it got if it was mere
changing of the default and the old behaviour was still available.
Even if the initial change was bogus in this regard as well then at
least one of X developers commenting on the bug could also say why the
patch fixing this aspect and adding the option to turn on the old
autodetection behaviour was not used.



More information about the xorg-devel mailing list