On Mon, Jan 02, 2023 at 08:33:05PM -0500, Thomas Dickey wrote: > On Mon, Jan 02, 2023 at 03:27:20PM -0500, Thomas Dickey wrote: > > On Mon, Jan 02, 2023 at 11:23:54AM +0000, Chris Green wrote: > > > So, after partially understanding the syntax highlighting issue I now > > > see that entering accented characters is still broken. > > > > > > As reported a while ago vile 9.8v gets confused when you try to enter > > > accented characters (I use a Multi_key, so to enter, for example an e > > > with an acute accent I hit Multi_key, e, '). > > > > > > On vile 9.8u (and earlier, I've checked 9.8t too) entering accented > > > characters with the Multi_key works perfectly. On 9.8v (as reported) > > > on entering an accented character an odd backslash sequence is echoed > > > and it's not the character entered. > > > > > > On vile 9.8w (from the Ubuntu repository) entering an accented > > > character does nothing if you enter just one accented character but if > > > you enter a second accented character that *does* get added. This > > > continues, entering a series of accented characters every other one > > > actually gets inserted. > > fwiw, this was introduced at the time that I fixed this bug: > > 20220805 (w) > + revise change for lins_chars() to handle UTF-8 in scripts, as that > interfered with normal insertions (report by Chris Green). > > and the "odd backslash" was from an earlier change to the same code: > > + modify lins_chars() to handle a case where a script inserts a UTF-8 > character (report by Thomas Dupond). > > I think I'll have time later this week to re-revisit this (and make both > scripts and interactive editing work at the same time).
see attached (9.8y will have other changes, but this is the small fix) -- Thomas E. Dickey <dic...@invisible-island.net> https://invisible-island.net
--- vile-9.8x+/map.c 2022-12-23 00:32:02.000000000 +0000
+++ vile-9.8y/map.c 2023-01-13 08:51:24.000000000 +0000
@@ -923,6 +926,18 @@
}
do {
+ /*
+ * This is a workaround for an ambiguity between the different return
+ * values from mapgetc(), on entry to this function:
+ * - If infloopcount is zero, mapgetc() has just called tgetc(), which
+ * in turn called sysmapped_c(), and then term.getc(). That pointer
+ * is initialized to vl_mb_getch(), which will decode UTF-8 into a
+ * Unicode value.
+ * - If infloopcount is nonzero, the function returns a byte (of UTF-8),
+ * which must be decoded later in this function.
+ */
+ int full_c = (infloopcount == 0); /* tgetc returned full char? */
+
(void) itb_init(&mappedchars, esc_c);
matched = maplookup(c,
@@ -949,7 +966,7 @@
c = mapgetc();
#if OPT_MULTIBYTE
- if (vl_encoding >= enc_UTF8 && b_is_utfXX(curbp)) {
+ if (remap && !full_c && (vl_encoding >= enc_UTF8 && b_is_utfXX(curbp))) {
char save[MAX_UTF8];
int have = 0;
int need = 9;
signature.asc
Description: PGP signature
