On my system, I need to remember my CUPS setup for every printer I use in multiple offices, departments and home. Most use duplex. None use letter size paper which is not available here and which has never been used by any printer used by this machine. Most are B&W.
Why? Because QT applications - unlike, for example, Adobe's acroread - ignore all default print settings set in CUPS.
For every job (note: every job - not every session or every initial print from a newly opened application - every single document) printed, I have to change:
- printer (defaulting to the default system printer is daft - should be last used but set that aside)
- QT default letter size to A4
- QT default of colour to B&W
- QT default single-sided to duplex
- other QT defaults as appropriate
This is despite the fact that everything is correct in CUPS. All printers default to A4, duplex, B&W etc.
It also makes it impossible to access advanced print functions for printers despite their being well-supported by the available drivers. Why? Because the QT options - and these are the only things which have any effect, the properties dialog having no effect on anything - are standard. Same options for all printers regardless of capabilities. It's a B&W printer? Never mind - QT still offers colour, even if CUPS knows all about it. Printer can use specially sized paper? Not if QT has anything to say about it, it won't! So even if you remember to change everything, there are lots of things you just cannot set up because the QT options are "one size fits all" and, oddly enough one size fits very, very few indeed.
The only case in which printing works as it should is if the defaults set in CUPS happen to match toe default options QT uses AND your printer limits itself to the capabilities QT thinks normal. Then I assume it would work. I've never been in that situation so it has never worked for me. It doesn't get the time right even twice a day.
Adobe can get this right - and they are hardly famed for their support of standard protocols or FOSS software - but it is, apparently, beyond the combined wit of Trolltech/Nokia, QT community developers, KDE developers, Red Hat, Fedora etc. Oh, wait! Actually, no. Many of these have actually fixed the issue. Several times. They have given QT patches. They have told QT repeatedly how to fix this - at least how to mitigate it.
But QT have looked this gift horse in the mouth. How much work can it really be to fix a problem for which other people have supplied pre-tested patches?