CUPS 1.6.3 fixes some compatibility issues with servers running CUPS 1.3.12 or older, fixes some general printing bugs, and fixes some minor security issues. Changes include:

• The configure script now prefers Clang over GCC.
• Fixed a compile problem on AIX (Issue #4307)
• The default IPP version did not always get set before creating a new IPP request message (<rdar://problem/14401718>)
• The lp, lpq, lpr, and lpstat now display an error message advising the use of the /version=1.1 ServerName option (<rdar://problem/14290628>)
• Added documentation about the /version=1.1 option to ServerName in client.conf (<rdar://problem/14216262>)
• httpStatus(HTTP_ERROR) did not return a useful error message (<rdar://problem/14217326>)
• The lp, lpq, lpr, and lpstat commands incorrectly ignored the default printer set in the lpoptions file (<rdar://problem/14216472>)
• Fixed a URI encoding issue for hostnames containing the ` (backquote) character (<rdar://problem/14243133>)
• Added support for RFC 6874’s IPv6 link local address format in URIs (<rdar://problem/13979453>)
• The USB backend could crash on libusb-based systems if USB was disabled in the BIOS (<rdar://problem/13875729>)
• Fixed a rounding error in the PWG media size mapping code (<rdar://problem/13493241>)
• Fixed several ipptool test files that used old STATUS names.
• Kerberos credentials could get truncated when printing to a shared printer.
• Printing using “ipps” URIs was not encrypted.
• Insecure ICC profiles prevented installation of user profiles for a printer on OS X.
• Added more USB quirks for the libusb-based backend (Issue #4311, <rdar://problem/13736470>)
• The Russian web interface templates were broken (Issue #4310)
• The scheduler no longer tries to do Kerberos authentication over the loopback interface.
• The IPP backend could fail to pause a job for authentication (Issue #4298)
• Fixed a regression on the handling of auth keys on OS X if the cups-files.conf was not present or did not contain a SystemAuthKey value.
• The scheduler incorrectly did a reverse lookup of the server address when HostNameLookups was turned off (Issue #4302)
• The scheduler incorrectly computed the final content type value when null filters were present.

Enjoy!