Hello,
zle_highlight array has fields like `fg_start_code' & `fg_end_code'.
The first is by default $'\x1b[', the second - 'm'.

ANSI colors and True-Colors are using those fields to construct and
emit the final, complete escape (by default) code to the terminal.

Yet the 248 colors are ignoring zle_highlight and are equipped with
terminal code by use of termcap. I think the following line (prompt.c:
~2044) does the conversion:

                tputs(tgoto(tcstr[tc], colour, colour), 1, putshout);

The patch makes the 256 colors to also use zle_highlight. This is the
main addition of the patch (also included the TrueColor case, to show
that this is done exactly this way):

    } else if (use_truecolor) {
        ptr += sprintf(ptr, "8;2;%d;%d;%d", colour >> 16,
                (colour >> 8) & 0xff, colour & 0xff);
+    } else if (colour > 7 && colour <= 255) {
+        ptr += sprintf(ptr, "8;5;%d", colour);

Is the patch acceptable? I imagine someone might suspect there exists
a terminal that uses escape codes for first 8 colors, but some other
kind of codes, collected in termcap, for remaining 248 colors. Can
this be considered impossible to accept the patch?

BTW. I've made X04 zle tests more rock-solid, because the debug prints
I added to track the 256-color/zle_highlight issue were slowing down
Zle, and a rare phenomenon (not occurred from the time I've simplified
Zle, disabled the echoing of input text) became very often: Ctrl-D was
appearing too quickly after Ctrl-A. A sleep of 333 ms made the problem
go away, and I think the issue is solved, and Zle tests are quite rock
solid now.
-- 
Sebastian Gniazdowski
News: https://twitter.com/ZdharmaI
IRC: https://kiwiirc.com/client/chat.freenode.net:+6697/#zplugin
Blog: http://zdharma.org