[argyllcms] Re: Argyll CMS in Fedora (and Mandriva)

  • From: "C." <ml-argyllcms712@xxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 21 Dec 2007 22:49:56 -0800

Thanks to Graeme Gill, Nicolas Mailhot, et. al.

I had just started using ArgyllCMS and was attempting to use it with the Huey 
Fedora 8 X86-64 when I came across the SRPM derived from the
packaging efforts of Nicolas Mailhot et. al.

I have successfully built RPMs from that package, though I had no
successful operation resulting from the files:

The regexp for generating the HID unbinding the Huey's hid/udev entry in:
...did not seem to correctly match the actual path corresponding to my USB Huey 
HID device,
so I had to manually unbind the Huey from HID control using the correct path I 
via ls -lR / dmesg et. al.

Also even after I had done that, I was unable to access the device as a normal
user due apparently to a permissions problem, though it did end up working as 
so I think there must be some other problems with the PolicyKit / SELINUX 
in the other files I mentioned.

Anyway I was glad to see the SRPM, the SPEC file, the patches for some of the 
buffer overflows, et. al.  And I found it all instructive / useful to help me 
get the
device working and save some time of mine trying to duplicate efforts in 
porting / packaging it.

It was of course quite nice that it ended up working with the Huey after a bit 
of effort, and
I've now generated my first profile with this configuration.

I think somewhere I have a note or reference listing what may be the correct 
form of the
HID udev unbinding regexp for the Huey; I think I saw it after an internet 
search on
some mailing list for another LINUX distribution talking about Huey / ArgyllCMS;
I'll post the details when I switch back to that system, find them, and test 
if they're of interest.  I have yet to find the information on the regexp
syntax and macro expansions available in the /etc subsystem implementing
argyllcms-0.70-96-huey-hid-workaround.rules, though it must be documented 
dmesg and ls -lR of the device tree all end up with somewhat similar very
cryptic nomenclatures for referring to the device.

Seems like it'd have been nice to include Vendor ID + Product ID in the path 
or at least similar nomenclature (maybe even the full path :) included in lsusb 
though, so one can more easily tell WHICH hardware device a given udev / hid 
refers  to in its 'cooked' pathname encoded format.

I had one other blocking problem in using ArgyllCMS's Fedora port that may be 
useful for
others trying to do this, and it might be relevant to include in a FAQ for LINUX
ArgyllCMS or some such thing.

I have been using the latest (169.04) beta drivers with the current update
version of xorg-X11 in Fedora 8.  I used a single monitor setup for the test,
and wasn't using TwinView et. al.  Everything seemed to be working OK with the
NVIDIA extensions, but I had no LUT access which prevented 
from working.  It turns out that all the extensions were present / functional
in the NVIDIA driver, though for some unknown reason I didn't have:
Load        "extmod"
in the appropriate (or any) section of my Xorg.conf.
This omission (non intuitively wrt. nomenclature) prevented Xv and
the XF86Video LUT access extensions that ArgyllCMS apparently depends
on from being functionally available and thus caused the aforementioned
ArgyllCMS utilities to fail.

It's a bit surprising that there isn't more explicit suggestion or configuration
to insert that configuration into Xorg since it seems like Nvidia's xorg 
doesn't put it in or suggest to do so when it's absent, and there's relatively
little mention of it anywhere online in the context of LUT access or ArgyllCMS 
that I noticed.

I also notice that there are some file conflicts between the NVIDIA driver
and the Fedora xorg-x11-server RPMs, and that using the software update
utility had caused some of the necessary NVIDIA files to get overwritten
because those files (e.g. libglx.so, et. al.) were listed as belonging to the
the xorg-x11-server RPM even though they'd been replaced by NVIDIA's ones
in the past.  I found / fixed that independently of my efforts with ArgyllCMS,
though, but I suspect it would also cause a fatal X11 extensions related
error in the inconsistent state if it was to occur on someone else's PC.

I'll look forward to seeing ArgyllCMS packages join Fedora and other LINUX
distributions, it's seemingly an excellent and useful addition!

I'd be glad to provide test feedback on future Beta ArgyllCMS
distributions and future alpha / beta SRPM packagings of it for Fedora
when those should become available.

Thanks to all who've been working on providing and packaging it!

Other related posts: