Description of problem: Pango's ligature detection is too keen (at least) with the Linux Libertine font. It picks up things like fi, fl and ffi mid-word, but also spots roman numerals and "No" as short for "number" when it's not appropriate. Version-Release number of selected component (if applicable): pango-1.16.4-2.fc7 linux-libertine-fonts-2.4.9-1.fc7 How reproducible: Always. Steps to Reproduce: 1. Start Pango-enabled app (e.g., Inkscape, Firefox). 2. Choose Linux Libertine font. 3. Find/type: difficult, UNIX, ASCII, Nobody. Actual results: See attached png (from Inkscape). Expected results: No ligature detection for UNIX, ASCII and Nobody.
Created attachment 256821 [details] Image displaying inappropriate ligatures
Pango is just doing what the font requests: make a ligature for those sequences. This is a bug in Libertine, so it should be fixed there.
Understood, component name and description changed accordingly.
Just relayed it to upstream. Roman numbers and "No" seem to be fixed with a more recent version available in Fedora 8. Unfortunately this version won't build on Fedora 7 because of an outdated fontforge package.
Confirmed --- fixed in 2.6.9-1.fc8.