client and allows it to be changed independently from the real active
pane stored in the window. This is can be used with session groups which
allow an independent current window (although it would be nice to have a
flag for this too and remove session groups). The client active pane is
only really useful interactively, many things (hooks, window-style,
zooming) still use the window active pane.
sessions it is wrong as often as it is right, and for windows in one
session it is pointless. Instead check TMUX_PANE to find the pane and
look for the MRU session as usual. GitHub issue 1793.
intended as the target client where the message should be displayed but
at some point (perhaps when -p was added), it was used for format
expansion too. This means it can get a bit weird where you have client
formats expanding for a client with a different current session than the
target session.
However, it is nice that display-message can be used to show information
about a specific client. So change so that the -c client will be used if
the session matches the target session (-t or default), otherwise the
best client will be chosen.
CMD_FIND_* flags in the cmd_entry and call it for the command. Commands
with special requirements call it themselves and update the target for
hooks to use.
do not look up windows as sessions (and panes as windows) when they are
qualified with a ':' or a '.'. So 'foo' as a window target will look for
windows and sessions called 'foo', but ':foo' will only look for
windows, and 'foo:' only for sessions. This means the common case of
using an unadorned session as a window target (send -tfoo) should
continue to work, but an explicit window will not get confused with a
session (send -t:foo).
falls back to an alternative if the tty name is not available. This is
clearer than overloading the client ttyname member and allows us to
remove the path stored in the tty struct, it should always be the same
as the client.
if the pane is zoomed, so instead add a new function to just check if
the pane is actually on screen (most commands still want to accept panes
invisible by zoom). Also reject panes outside the window for various
special targets. Problem reported by Sean Haugh.
but there is also now a global command queue. Instead of command queues
being dispatched on demand from wherever the command happens to be
added, they are now all dispatched from the top level server
loop. Command queues may now also include callbacks as well as commands,
and items may be inserted after the current command as well as at the end.
This all makes command queues significantly more predictable and easier
to use, and avoids the complex multiple nested command queues used by
source-file, if-shell and friends.
A mass rename of struct cmdq to a better name (cmdq_item probably) is
coming.
- Prepare the state again before the "after" hooks are run, because the
command may have killed or moved windows.
- Use the hooks list from the newly prepared target, not the old hooks
list (only matters for new-session really).
- Correctly detect an invalid current state and ignore it in
cmd_find_target ("killw; swapw").
- Change neww, new, killp, killw, splitw, swapp, swapw to update the
current state (used if no explicit target is given) to something more
useful after they have finished. For example, neww changes it to the
newly created window.
Hooks are still relatively new and primitive so there are likely to be
more changes to come.
Parts based on bug reports from Uwe Werler and Iblis Lin.
