[Openicc] meta data in test chart

[Robert Krawitz]

On Fri, 21 Jan 2011 08:09:33 -0700, Chris Murphy wrote:
> On Jan 20, 2011, at 10:00 PM, Kai-Uwe Behrmann wrote:
>> I think we have to do a cut here. IMO it would be straight forward
>> to just care about settings being colour related. After all that
>> sounds to me like what you want with the calibration "recipe".
>
> I think restricted the shareable package to that of just strict
> color related settings is a mistake. If someone is using a thick
> media, then paper feed or platen adjustments are obviously necessary
> to properly reproduce the proper printing condition. If such
> settings are straight out not in scope for embedding in the output
> device profile, then I'd recommend a different approach, where are
> package file format that includes the ICC profiles and all print
> settings is used, rather that stuffing them into an ICC profile.

Agreed.  I personally think that things like that should be in scope.
The recipe should be "everything that you need in order to match the
output", and if the platen setting isn't right, all kinds of things
could go wrong.  If the paper jams, you certainly won't match the
output!

Maybe we should turn this around a bit and discuss what kinds of
things are *not* in scope here, and I suspect we'll come up with a
rather short list.  Looking at the Gutenprint source, I can identify
the following that I think people can reasonably expect to change
without any effect on printing:

* Dimension-related settings:
  - Page size
  - CD inner and outer radius (I could see counter-arguments here,
    particularly for the inner radius)

* Paper handling (note that many paper handling options are in scope):
  - Input slot
  - Duplex
  - Full bleed

But then we come to the question of whether someone should be able to
use a "recipe" and still make selected changes based on their own fine
tuning.  I think this should be possible, but it should be very clear
what is being changed.

> And really, a particular set of settings is only valid for a
> particular PPD. Maybe nothing changes as the PPD version changes,
> but perhaps something does change that relates to color and then the
> settings aren't valid. So again I think the package that contains
> PPD, settings, ICC profile, calibration, is a better way to
> transport and share. It's easier, more intuitive, more
> reliable. Otherwise you get situations where users have the correct
> settings, different driver version, don't have a calibration file,
> and now they can't actually reproduce the printing conditions.

Gutenprint locks PPD file versions to Gutenprint versions for
basically this reason; you can't use 5.2.5 PPD files with 5.2.6 (the
actual reason is that allowable option values may change and we don't
want to pass invalid values; they do get checked, but the result would
be a failure to print if there were inconsistencies).  We do provide
an upgrade tool to permit version upgrades.

We also generally don't change output for non-experimental printers
during the life of a minor release (e. g. 5.2.x).  We're not absolute
about this; for example, I've checked in some big improvements for the
Epson Stylus Pro x880 and am waiting on some tests to decide whether
to commit them for the x800 printers, but we may also bump the minor
version.  Incidentally, though, it would be possible to match the
current output with Gutenprint 5.2.6 by means of options in a "recipe".