Mercurial > hg > xemacs-beta
diff src/events.h @ 5803:b79e1e02bf01
Preserve extent information in the command builder code.
src/ChangeLog addition:
2014-07-14 Aidan Kehoe <kehoea@parhasard.net>
* event-stream.c:
* event-stream.c (mark_command_builder):
* event-stream.c (finalize_command_builder): Removed.
* event-stream.c (allocate_command_builder):
* event-stream.c (free_command_builder): Removed. Use
free_normal_lisp_object() instead.
* event-stream.c (echo_key_event):
* event-stream.c (regenerate_echo_keys_from_this_command_keys):
Detach all extents here.
* event-stream.c (maybe_echo_keys):
* event-stream.c (reset_key_echo):
* event-stream.c (execute_help_form):
* event-stream.c (Fnext_event):
* event-stream.c (command_builder_find_leaf_no_jit_binding):
* event-stream.c (command_builder_find_leaf):
* event-stream.c (lookup_command_event):
* events.h (struct command_builder):
Move the command builder's echo_buf to being a Lisp string rather
than a malloced Ibyte array. This allows passing through extent
information, which was previously dropped. It also simplifies the
allocation and release code for the command builder.
Rename echo_buf_index to echo_buf_fill_pointer, better reflecting
its function.
Don't rely on zero-termination (something not particularly
compatible with Lisp-level code) when showing a substring of
echo_buf that differs from that designated by
echo_buf_fill_pointer, keep a separate counter instead and use
that.
* minibuf.c:
* minibuf.c (echo_area_append):
Use the new START and END keyword arguments to #'append-message,
rather than consing a new string for basically every #'next-event
prompt displayed.
test/ChangeLog addition:
2014-07-14 Aidan Kehoe <kehoea@parhasard.net>
* automated/extent-tests.el:
Check that extent information is passed through to the echo area
correctly with #'next-event's PROMPT argument.
lisp/ChangeLog addition:
2014-07-14 Aidan Kehoe <kehoea@parhasard.net>
* simple.el (raw-append-message):
Use #'write-sequence in this, take its START and END keyword
arguments, so our callers don't have to cons as much.
* simple.el (append-message):
Pass through START and END here.
author | Aidan Kehoe <kehoea@parhasard.net> |
---|---|
date | Mon, 14 Jul 2014 13:42:42 +0100 |
parents | 8d29f1c4bb98 |
children | a216b3c2b09e |
line wrap: on
line diff
--- a/src/events.h Wed Jul 02 17:45:49 2014 +0100 +++ b/src/events.h Mon Jul 14 13:42:42 2014 +0100 @@ -1148,11 +1148,13 @@ translation loop). If this is nil, then the next-read event is the first that can begin a function key sequence. */ Lisp_Object first_mungeable_event[2]; - Ibyte *echo_buf; - - Bytecount echo_buf_length; /* size of echo_buf */ - Bytecount echo_buf_index; /* index into echo_buf + Lisp_Object echo_buf; + Bytecount echo_buf_fill_pointer; /* Fill pointer for echo_buf. * -1 before doing echoing for new cmd */ + Bytecount echo_buf_end; /* End of the text to be shown in + echo_buf. Can be after the fill + pointer, but usually identical to + it */ /* Self-insert-command is magic in that it doesn't always push an undo- boundary: up to 20 consecutive self-inserts can happen before an undo- boundary is pushed. This variable is that counter.