[colord] Avoiding double correction in graphical apps

[Richard Hughes]

On 9 December 2013 14:25, Alexey Galakhov <agalakhov at gmail.com> wrote:
> Of course one may give sRGB output profile
> to that software but this is unsuitable in case of wide gamut monitors.

Right. Personally I'd given up on X as it's basically going to be a
hack-on-hack without a completely new extension to the protocol. The
correct way to do this is easy with wayland. We can tag a surface (or
sub-surface) with an ICC profile and let the compositor handle the
blending and format conversion.

> http://www.oyranos.org/wiki/index.php?title=Oyranos/net-color_spec_0.2
> to do that. Let's support it?

No way. That's IPC using X atoms, which isn't going to work when each
legacy app is in it's own X server instance. Using wayland we can do
this much better, and I've written quite a lot on the wayland mailing
list about that already (and wrote the cms-colord code :)

> A similar problem exists in ghostscript+gutenprint pipeline (with or
> without CUPS).

No, the cups pipelines query colord and don't apply a color profile
when calibrating. I agree we need to add the PDF attribute for network
printers, but it already works well calibrating locally for me on
Fedora 19.

Richard