[Openicc] Re: xcalib in Oyranos package

[Stefan Döhla]

Kai-Uwe, list
> what does you think about including xcalib into an Oyranos package? It 
> would probably make life easier for distributors and of course Oyranos
> more useable :-)
>   
FreeBSD comes with xcalib as an optional package already. I don't care 
what people do with xcalib as long as they respect the license (which is 
currently rather strict). However, I release new versions once a while 
and being up-to-date is not always easy if there are many repositories. 
If someone feels, that he needs a customized version of xcalib, feel 
free. However, I would appreciate to be informed about the changes and 
fixes if others could benefit.
Additionally, it maybe an issue of the used parsers/LUT-APIs: Currently 
there are three parsers and three LUT-APIs:
- Parsers:
-- own (very simple, not absolutely robust, supports this mLUT thing of 
ProfileMechanic, includes gamma limit handling - the one I work with 
most of the time)
-- icclib by Graeme Gill (stable, bundled with the .tar.gz file of the 
xcalib sources)
-- patched LCMS by Marti Maria (this was more a proof-of-concept - uses 
LCMS to smooth the ramps - don't know about Marti's opinion)
- LUT-APIs:
-- X11 XVidMode extension (the normal one)
-- Win32 (yes, this works! and it works good. it can even be used as a 
program in the StartUp folder since Win32 has an obtainable default 
profile for the display)
-- ATI fglrx (this is for the proprietary ATI driver together with X11)

Not all combinations are right now possible, because of my laziness to 
modularize, but this shows, that the core is the same for all these 
different platforms but for users it's quite complicated. For now, I 
trust the experienced user to select the right options (or patching the 
Makefile) and giving me feedback about problems.
> By the way, you are allready ahead in multi monitor support?
Multi head support is already possible by using the standard X 
interfaces and running xcalib for every display. The limit is usually 
Xinerama, which doesn't allow to set multiple LUTs, since it's one 
logical screen. The other limit is the hardware itself, which has only 
one LUT for either the primary output or common for all outputs. I don't 
know about the newer cards (especially ATI, nVidia Quadro series and the 
Matrox Parhelia), whether they can be set independently, but there are 
probably others, who can jump in and report. Did I forget to mention 
calibration with the monitor's LUT? But that's maybe another story ...

>  Would it be a 
> good time to enhence the ICC_PROFILE spec about your needs? Then you could 
> fetch the profile directly from the X server (as Oyranos would).
>   
I didn't feel very comfortable when it came to the mentioned spec. But I 
will support it in some way in the near future. It should be possible to 
set the profile and LUT with a single command. The same applies to 
default profiles: they should be easily settable (maybe in the main 
config or somewhere else), so that the LUTs can be changed 
semi-automatically.

Stefan