aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorDaniel Eklöf <daniel@ekloef.se>2021-05-10 15:57:14 +0000
committerDaniel Eklöf <daniel@ekloef.se>2021-05-10 15:57:14 +0000
commit2e8bea0a5d69a82f38b0b9e88ca292c84ddf2852 (patch)
tree35f650ca3582e758940bfb751f6958710e3e990e
parentMerge branch 'sixel-assertion' (diff)
wayland: add comment describing the configure preempt workaround
-rw-r--r--wayland.c27
1 files changed, 26 insertions, 1 deletions
diff --git a/wayland.c b/wayland.c
index 8985e119..ac43d96f 100644
--- a/wayland.c
+++ b/wayland.c
@@ -664,6 +664,30 @@ xdg_surface_configure(void *data, struct xdg_surface *xdg_surface,
xdg_surface_ack_configure(xdg_surface, serial);
+#if 1
+ /*
+ * This was done as a workaround for a bug in Sway
+ * (https://github.com/swaywm/sway/issues/6023
+ *
+ * Sway waited for configure ACKs from hidden windows. These never
+ * arrived because Sway never scheduled any frame callbacks for
+ * these hidden windows...
+ *
+ * This has bene fixed as of Sway-1.6.
+ *
+ * Unfortunately, slow moving distros (Debian *cough* are still
+ * stuck on Sway-1.5, and will be so for the foreseeable
+ * future.
+ *
+ * TODO: check for e.g. SWAYSOCK and only pre-empt if it exists?
+ * We'd still be pre-emptying on Sway-1.6, but at least other
+ * compositors would be unaffected.
+ *
+ * Or do we ignore Sway-1.5 users and point them to the Sway bug?
+ *
+ * Note that it doesn't appear to cause any issues to have this
+ * enabled, on any compositor.
+ */
if (term->window->frame_callback != NULL) {
/*
* Preempt render scheduling.
@@ -676,9 +700,10 @@ xdg_surface_configure(void *data, struct xdg_surface *xdg_surface,
term->window->frame_callback = NULL;
}
+#endif
#if 1
/*
- * TODO: decide if we should to the last “forced” call when ending
+ * TODO: decide if we should do the last “forced” call when ending
* an interactive resize.
*
* Without it, the last TIOCSWINSZ sent to the client will be a