From: Nathaniel Wesley Filardo <nwf@cs.jhu.edu>
Date: Sat, 29 Jul 2017 19:32:45 +0000 (-0400)
Subject: tq bugfixes (and API change)
X-Git-Url: https://hydra-www.ietfng.org/gitweb/?a=commitdiff_plain;h=48e6ed5e2042cfc24d8899d4a8838fd74d1814eb;p=acmetensortoys-esp-lua_lamp

tq bugfixes (and API change)

- not all lapsed time should be credited to the head node
- needs to reset the time base (_tst) in all cases
- always return the object for cancellation
- defend against time going backwards (just in case)

While here, rework to simplify, simplify API, and add comments
---

diff --git a/linux-draw.lua b/linux-draw.lua
index cceffe1..e9cb104 100644
--- a/linux-draw.lua
+++ b/linux-draw.lua
@@ -29,10 +29,10 @@ end
 tq = dofile("tq/tq.lua")(nil)
 tq.__emu_lastcb = 0
 tq.now = function() return cq.monotime() * 1000000 end
-tq.arm = function(self,fn,t)
+tq.arm = function(self,t,et)
   local cbix = tq.__emu_lastcb + 1
   tq.__emu_lastcb = cbix
-  cqc:wrap(function() cq.poll(t/1000) ; if tq.__emu_lastcb == cbix then fn() end end)
+  cqc:wrap(function() cq.poll(t/1000) ; if tq.__emu_lastcb == cbix then tq:fire() end end)
 end
 
 -- how backwards is this!?  We are using tq as faked above for tmr support