Bugs item #481, was opened at 2006-02-13 18:04 You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=481&group_id=106 Category: hz Group: v1.40.0 Status: Open Resolution: None Priority: 5 Submitted By: Robert (schlcht) Assigned to: The Thanh Han (hanthethanh) Summary: "This can't happen (flushing)" with extra kerning+expansion+ligatures Initial Comment: With the following test file, pdftex will abort with "This can't happen (flushing)": \pdfadjustspacing=1 \pdffontexpand\font 20 20 4 autoexpand \pdfappendkern=1 \knaccode\font`i=100 \hsize10em foo fii foo foo foo foo % but this works: %foo f{}ii foo foo foo foo \bye Hartmut's analysis was: ----------------------------------- FWIW, this happens in function do_subst_font(). But it starts with auto_expand_kern(), where the given kern indeed is appended to the i of "fi" and also to the 2nd "i". It seems as if at this time "fi" is not yet seen as a ligature. Then in do_subst_font() the "fi" is somehow already a ligature, and now the kern is part of the lig_ptr list (which is wrong). Then font(r) is recursively changed to 51, the expanded font number, and so the kern, which should not be in this list, gets the 51 as its type. But it must be 11 for a kern. Maybe a change should be that the kern does not end up in the lig_ptr list, but should come after the "fi" ligature. No idea, how. ----------------------------------- ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=481&group_id=106