Hi
I think a better solution might be if control clients do not affect the
session and window size until they call refresh-client -C - so they
effectively have no size at all until then. This means control clients
that don't care about the output, perhaps if they are just looking for
certain events, can avoid changing the size at all.
How about this?
diff --git a/cmd-refresh-client.c b/cmd-refresh-client.c
index 5190df89..6af3362b 100644
--- a/cmd-refresh-client.c
+++ b/cmd-refresh-client.c
@@ -67,8 +67,10 @@ cmd_refresh_client_exec(struct cmd *self, struct cmdq_item
*item)
cmdq_error(item, "not a control client");
return (CMD_RETURN_ERROR);
}
- if (tty_set_size(&c->tty, w, h))
+ if (tty_set_size(&c->tty, w, h)) {
+ c->flags |= CLIENT_SIZECHANGED;
recalculate_sizes();
+ }
} else if (args_has(args, 'S')) {
c->flags |= CLIENT_STATUSFORCE;
server_status_client(c);
diff --git a/resize.c b/resize.c
index ff1c9eec..4c41f769 100644
--- a/resize.c
+++ b/resize.c
@@ -60,6 +60,9 @@ recalculate_sizes(void)
TAILQ_FOREACH(c, &clients, entry) {
if (c->flags & CLIENT_SUSPENDED)
continue;
+ if ((c->flags & (CLIENT_CONTROL|CLIENT_SIZECHANGED)) ==
+ CLIENT_CONTROL)
+ continue;
if (c->session == s) {
if (c->tty.sx < ssx)
ssx = c->tty.sx;
diff --git a/tmux.h b/tmux.h
index e8ad55ff..3edb5457 100644
--- a/tmux.h
+++ b/tmux.h
@@ -1349,6 +1349,7 @@ struct client {
#define CLIENT_STATUSFORCE 0x80000
#define CLIENT_DOUBLECLICK 0x100000
#define CLIENT_TRIPLECLICK 0x200000
+#define CLIENT_SIZECHANGED 0x400000
int flags;
struct key_table *keytable;
On Tue, May 09, 2017 at 11:20:53PM -0700, George Nachman wrote:
> When a control mode client attaches, the client size is unknown. A control
> mode client's tty is initialized to 80x24. When attaching,
> recalculate_sizes gets called, which changes pane layouts.
> Unlike regular clients, control mode clients should always set the client
> size explicitly with refresh-client -C, but they don't get a chance to do
> so early enough.
> To avoid corrupting the existing layouts when detaching and reattaching,
> this patch sets a control mode client's tty size when attaching for the
> first time based on the existing window sizes in the session.
> One issue: Behavior in aggressive-resize isn't great. I'm not sure how
> much it matters in practice. iTerm2 refuses to attach when
> aggressive-resize is enabled.
>
> --
> You received this message because you are subscribed to the Google Groups
> "tmux-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To post to this group, send email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
> diff --git a/cmd-attach-session.c b/cmd-attach-session.c
> index 81ed4a8..4cc84c7 100644
> --- a/cmd-attach-session.c
> +++ b/cmd-attach-session.c
> @@ -58,6 +58,8 @@ cmd_attach_session(struct cmdq_item *item, const char
> *tflag, int dflag,
> struct winlink *wl;
> struct window_pane *wp;
> char *cause;
> + int of = c->flags;
> + u_int sx, sy;
>
> if (RB_EMPTY(&sessions)) {
> cmdq_error(item, "no sessions");
> @@ -153,6 +155,16 @@ cmd_attach_session(struct cmdq_item *item, const char
> *tflag, int dflag,
> notify_client("client-attached", c);
> c->flags |= CLIENT_ATTACHED;
> }
> + if (!(of & CLIENT_ATTACHED) && (c->flags & CLIENT_CONTROL)) {
> + sx = sy = 0;
> + RB_FOREACH(wl, winlinks, &s->windows) {
> + if (wl->window->sx > sx) sx = wl->window->sx;
> + if (wl->window->sy > sy) sy = wl->window->sy;
> + }
> + if (sx > 0 && sy > 0) {
> + tty_set_size(&c->tty, sx, sy);
> + }
> + }
> recalculate_sizes();
> alerts_check_session(s);
> server_update_socket();
--
You received this message because you are subscribed to the Google Groups
"tmux-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send an email to [email protected].
For more options, visit https://groups.google.com/d/optout.
