summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorNicholas Marriott <nicholas.marriott@gmail.com>2017-04-20 15:40:28 +0100
committerNicholas Marriott <nicholas.marriott@gmail.com>2017-04-20 15:40:28 +0100
commit7934e7a6b518ec96d95fc4e9303d6fdafcac54bc (patch)
tree5f33fdf809b2ee1e5b70435b9f8a367d4ea61b5c
parentaebbb15fe2d1ef900e7a66dc16bdf89f2a2394d6 (diff)
Notes on target resolution.
-rw-r--r--TODO24
1 files changed, 24 insertions, 0 deletions
diff --git a/TODO b/TODO
index fba405f1..8bde3238 100644
--- a/TODO
+++ b/TODO
@@ -120,3 +120,27 @@
* multiple hooks with the same name?
* finish hooks for notifys
* for session_closed, if no sessions at all, perhaps fake up a temporary one
+
+- current target resolution is still confusing and probably wrong:
+ * what if I
+ bind -n DoubleClick1Pane selectp -U\; resizep -Z
+ and click on a pane that isn't the active pane? It should make the
+ pane above the one I clicked on active and zoom it
+ * bind q select-pane -U\; resize-pane -Z should do the obvious
+ * bind q neww\; splitw as well
+ * should we always start with a "current target" and commands can
+ update it?
+ * a few cases:
+ 1) normal key bindings - fine, we always know the full target
+ but it is important that commands like neww and selectp
+ use the updated active window or pane;
+ 2) mouse key binding - same, but needs to be the mouse target;
+ 3) from a new client ("tmux whatever") - we may know the best
+ s,wl,wp from the pty, guess the client, or guess all if
+ we don't know;
+ 4) hooks - we may know or we may not;
+ 5) control mode - same as a key binding really, and not
+ specifying a target is stupid anyway;
+ 6) if-shell/run-shell - may be able to preserve its target,
+ else guess.
+ * of course -t and -s override the current target as now.