perf(dashboard): draw at 0.1s only while the dashboard is visible

Conky gives Lua no way to request one redraw, so a scroll showed up to
2s late. Pointer enter and leave, which Hyprland sends on every toggle
of the special workspace, now switch conky_set_update_interval between
0.1s and the config's 2s.

At that rate the CPU and network counters would sample over 0.1s
windows and the chart would fill 20x faster, so both now sample through
data.new_gate on the /proc/uptime clock, about every 2s, and redraw
their last values in between. The samplers move to texeci: execi runs
in the main loop and a 0.7s nvtable or an Immich download stalled the
scroll.

Measured: about 12% of one core while visible, 0.5% while hidden.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

1 files changed, 12 insertions, 4 deletions

diff --git a/widgets/network.lua b/widgets/network.lua
index 3a11521..20b0573 100644
--- a/widgets/network.lua
+++ b/widgets/network.lua
@@ -37,6 +37,11 @@ local rx_hist, tx_hist = {}, {}
 local hist_len = 0
 local lan_addr = nil
 
+-- Sampled every ~2s whatever the draw rate, so the chart keeps one point per
+-- sample rather than per draw. 1.9 not 2: draws jitter by a few ms.
+local gate = data.new_gate(1.9)
+local down, up = nil, nil
+
 -- Append to a fixed-length history, dropping the oldest. A plain array with
 -- table.remove(1) rather than a circular buffer with an index: the lengths
 -- here are a few hundred at most and the draw is every two seconds, so the
@@ -73,8 +78,11 @@ function M.draw(cr, rect, colors)
   rx_counter = rx_counter or data.new_rate_counter(interval)
   tx_counter = tx_counter or data.new_rate_counter(interval)
 
-  local down = rx_counter:sample(read_counter('rx_bytes'))
-  local up   = tx_counter:sample(read_counter('tx_bytes'))
+  local sampled, now = gate()
+  if sampled then
+    down = rx_counter:sample(read_counter('rx_bytes'), now)
+    up   = tx_counter:sample(read_counter('tx_bytes'), now)
+  end
 
   -- Fluid type, as every other card does: one base size S drives everything,
   -- taken as the smaller of a height budget and a measured width fit.
@@ -140,8 +148,8 @@ function M.draw(cr, rect, colors)
       while #rx_hist > hist_len do table.remove(rx_hist, 1) end
       while #tx_hist > hist_len do table.remove(tx_hist, 1) end
     end
-    if down then push(rx_hist, down, hist_len) end
-    if up then push(tx_hist, up, hist_len) end
+    if sampled and down then push(rx_hist, down, hist_len) end
+    if sampled and up then push(tx_hist, up, hist_len) end
 
     -- One ceiling for both series. Scaled independently they would lie about
     -- their relative size, which is the entire reason to draw them together.