[Openicc] XICC specification draft

[Craig Ringer]

On Thu, 2005-06-23 at 12:09 +0100, Ross Burton wrote:
> Hi,
> 
> I've documented the atom (_ICC_PROFILE) my xicc tool and eog patches use
> to set/get the ICC profile for an X screen, and have put a draft of it
> here:
> 
> http://burtonini.com/temp/xicc-specification-0.1.html
> 
> I'd appreciate it if people would review it.

Looks clear enough to me.

I'd suggest saying that the atom doesn't have to exist, and if it does
not the display should be assumed to be uncalibrated. If no display
profile is explicitly configured, the atom should be unset.

Stating the obvious, but you never know ... someone writing some config
tool might well say "or, if I don't know what profile to use, I'll just
set SRGB".

http://bugs.scribus.net/view.php?id=2117

One minor issue: "As profiles can be large applications should read the
profile once, and cache it." Might that not become a significant memory
cost if ICC support is integrated into toolkits etc, so every tiny
application ends up reading and caching the profile? On the flipside
there's bandwidth and context switches, so I don't know the right
answer, but I thought I'd bring that up as a possible issue.

--
Craig Ringer