[cups.general] Why not "recommend" PPDs in the NickName?

[Johannes Meixner]

Hello,

first of all:  There is no need to shout.

On Jan 29 09:50 Michael Sweet wrote (shortened):
> Johannes Meixner wrote:
> > But meanwhile I think probabilities are wrong because
> > of their limited range.
> > A driver which is considered right now to produce perfect
> > printout quality would have the probability 1 regarding quality
> > but how should we do the ranking for a new driver which produces
> > even better quality (it cannot have a probability of 1.2).
> 
> Will the same printer suddenly inherit new features and capabilities?

When you read carefully you will find out that I didn't talk
about new features but about a new driver for the same printer.

For example DeskJet printers before and after HPIJS:
Before HPIJS the Ghostscript PCL3 drivers (like cdj*)
may have beed considered to be 100% but since HPIJS
they should be downgraded to e.g. 80%. but downgrading
is problematic so that upgrade HPIJS to 120% would be easier.
This applies both for quality and for speed because
as far as I remember the Ghostscript cdj* drivers neither
support the built-in dithering nor the draft mode in certain
DeskJet printers.


> WE ARE NOT MEASURING QUALITY!
> 
> Quality is subjective and can change over time.  Instead, we
> want to recommend a driver that a) supports the most features/
> capabilities of a printer and b) provides the fastest printing
> for a particular use case.

Then the rating is not very useful for many users because
the printout quality is also very important for many users.

For text-only printing speed is perhaps most important but
e.g. for photo printing speed is less important than quality
so that a speed-only rating would recommend the wrong driver
for photo printing if there are two drivers which support it
is but one is faster with less quality than the other.

A not very useful rating system may be somehow circumvented
or ignored by this or that printer setup tool to choose the
really "best" PPD by default.


> Don't ship alpha-quality software that doesn't work on all of your
> supported platforms!

Very many thanks to remind us what we must do to make
as many as possible of our users as happy as possible
(which does of course not mean that it is possible
to provide perfect results for all users).


Kind Regards
Johannes Meixner
-- 
SUSE LINUX Products GmbH, Maxfeldstrasse 5      Mail: jsmeix at suse.de
90409 Nuernberg, Germany                    WWW: http://www.suse.de/