> ... Why does GtkEntry work correctly? Because there the inserted text goes through lots of other code than just directly mapping from key presses to characters, which is never going to work in general. You need to read up on GtkIMContext, I think. > But because the GtkEntry handles this correctly I'm > convinced there is some function (perhaps hidden) in Gt(d)k > which has the knowledge of all keyboards layouts in all languages. Not really knowledge of keyboard layouts. It's a table of compose sequences, check gtkimcontextsimple.c and gtkimcontextsimpleseqs.h, and heuristics to handle dead diacritics, see check_algorithmically() in gtkimcontextsimple.c. > But actually a text editor user would expect the letters appear > as soon as he/she presses a key. Actually, no. When the user presses a dead diacritic key, nothing should be visible or inserted in an editor buffer. Only after the next key can the code know what should be inserted. You don't want a dead diacritic key (like [~]) to cause the corresponding spacing diacritic (<~>) (sorry for notation I just invented... in the first case I mean a key, in the second a Unicode code point) to be inserted, and when the user then types [n] the <~> then be replaced with <ñ> . Well, a text editor could behave like that, but it would be different then from how dead diacritics normally behave. --tml _______________________________________________ gtk-list mailing list gtk-list@xxxxxxxxx http://mail.gnome.org/mailman/listinfo/gtk-list
