[Owen Taylor] > On Fri, 2005-02-25 at 16:19 +0100, Haakon Riiser wrote: >> Just using the font as if it were ISO10646-1 would be good >> enough. Should't that be a trivial patch? It is, at least, >> much better than the current behavior, which I don't see any >> use for. > To this I can only reply: > > Noooooooooooooooooooooooooooooooo. > > If an input stream has a ? in it, then that should come out as > a ? or nothing. It should never come out as a ???. > > Misencoding input text in order to match a misencoded (or > misinterpreted) font is one of the ultimate evils of the font > and text handling world. I don't see it as a problem, but OK. Here's another idea: how about making it possible to define aliases for font encodings in fonts.conf? Then I could tell fontconfig that IBM cp437 is another name for ISO8859-1; there would be no side effect for people who don't specifically ask for it, and everyone is happy. :-) -- Haakon
