summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorJason Rumney <jasonr@gnu.org>2003-08-31 22:22:29 +0000
committerJason Rumney <jasonr@gnu.org>2003-08-31 22:22:29 +0000
commite9eee23e17d22e1def722b9c5f2779ac44db6ed9 (patch)
tree5cb330d59377fdfc838e5ce3ee56b8ff695a5368
parent4bc2315e9ee39791abb683a4ba4b27e7a69d862a (diff)
downloademacs-e9eee23e17d22e1def722b9c5f2779ac44db6ed9.tar.gz
Document last change.
-rw-r--r--src/w32term.c11
1 files changed, 11 insertions, 0 deletions
diff --git a/src/w32term.c b/src/w32term.c
index 16e0f4667c0..7531dd0fd05 100644
--- a/src/w32term.c
+++ b/src/w32term.c
@@ -896,6 +896,17 @@ w32_per_char_metric (font, char2b, font_type)
xassert (font && char2b);
+ /* TODO: This function is currently called through the RIF, and in
+ some cases font_type is UNKNOWN_FONT. We currently allow the
+ cached metrics to be used, which seems to work, but in cases
+ where font_type is UNKNOWN_FONT, we probably haven't encoded
+ char2b appropriately. All callers need checking to see what they
+ are passing. This is most likely to affect variable width fonts
+ outside the Latin-1 range, particularly in languages like Thai
+ that rely on rbearing and lbearing to provide composition. I
+ don't think that is working currently anyway, but we don't seem
+ to have anyone testing such languages on Windows. */
+
/* Handle the common cases quickly. */
if (!font->bdf && font->per_char == NULL)
/* TODO: determine whether char2b exists in font? */