index : advtrains.git	atcjit doc l10n luaatcdebug master profiler timetable tracy trainhud trainhud-unifont
	Advtrains mod for minetest.	orwell

aboutsummaryrefslogtreecommitdiff

log msg author committer range

path: root/serialize_lib

	Commit message (Collapse)	Author	Age
*	Remove non-deterministic test	Gabriel Pérez-Cerezo	2021-11-10
|
*	Activate serialize_lib unit tests	Gabriel Pérez-Cerezo	2021-11-10
|
*	Merge commit 'd4593491efbcab47efd918d7918b97b4621828b3'	orwell96	2021-05-03
|
*	Update serialize_lib	orwell96	2021-04-04
|
*	Add 'serialize_lib/' from commit 'a6e8b8b4353863ad563a4d5187f40fea702ea2de'	orwell96	2021-03-13
| | | | | |	git-subtree-dir: serialize_lib git-subtree-mainline: b82e10051d69730b7459cceae4e4e8719b4445d0 git-subtree-split: a6e8b8b4353863ad563a4d5187f40fea702ea2de
*	Remove serialize_lib in order to re-add it as subtree	orwell96	2021-03-13
|
*	add unit tests for serialize_lib	Gabriel Pérez-Cerezo	2021-01-31
|
*	Fix serialisation: breach of contract, file left open	Blockhead	2021-01-18
| | | |	Previous commit did not fix saving, but is kept because there is a corner case for which it is required (see MT forum)
*	Fix atomic saving on windows when save file does not exist	orwell96	2021-01-17
|
*	Some more serializer fixes (backported from new_lzb):	orwell96	2021-01-12
| | | | | |	- Move DUMP_DEBUG_SAVE block before the actual saving so it can be used to trace serializer errors - Don't crash on functions in data, ignore them silently - Increase the save interval
*	serialize_lib: Allow empty strings in key	orwell96	2021-01-12
|
*	Serialize_lib: finish up and add atomic api	orwell96	2021-01-12
|
*	Implement basic serialization and file opening	orwell96	2021-01-12

generated by cgit v1.2.3 (git 2.39.1) at 2025-07-10 13:06:25 +0000

-- Setting and clearing routes

-- TODO duplicate
local lntrans = { "A", "B" }
local function sigd_to_string(sigd)
	return minetest.pos_to_string(sigd.p).." / "..lntrans[sigd.s]
end

local ildb = advtrains.interlocking.db
local ilrs = {}

local sigd_equal = advtrains.interlocking.sigd_equal

-- table containing locked points
-- also manual locks (maintenance a.s.o.) are recorded here
-- [pts] = { 
--		[n] = { [by = <ts_id>], rsn = <human-readable text>, [origin = <sigd>] }
--	}
ilrs.rte_locks = {}
ilrs.rte_callbacks = {
	ts = {},
	lck = {}
}


-- main route setting. First checks if everything can be set as designated,
-- then (if "try" is not set) actually sets it
-- returns:
-- true - route can be/was successfully set
-- false, message, cbts, cblk - something went wrong, what is contained in the message.
-- cbts: the ts id of the conflicting ts, cblk: the pts of the conflicting component
function ilrs.set_route(signal, route, try)
	if not try then
		local tsuc, trsn, cbts, cblk = ilrs.set_route(signal, route, true)
		if not tsuc then
			return false, trsn, cbts, cblk
		end
	end

	
	-- we start at the tc designated by signal
	local c_sigd = signal
	local first = true
	local i = 1
	local rtename = route.name
	local signalname = ildb.get_tcbs(signal).signal_name
	local c_tcbs, c_ts_id, c_ts, c_rseg, c_lckp
	while c_sigd and i<=#route do
		c_tcbs = ildb.get_tcbs(c_sigd)
		if not c_tcbs then
			if not try then atwarn("Did not find TCBS",c_sigd,"while setting route",rtename,"of",signal) end
			return false, "No TCB found at "..sigd_to_string(c_sigd)..". Please reconfigure route!"
		end
		c_ts_id = c_tcbs.ts_id
		if not c_ts_id then
			if not try then atwarn("Encountered End-Of-Interlocking while setting route",rtename,"of",signal) end
			return false, "No track section adjacent to "..sigd_to_string(c_sigd)..". Please reconfigure route!"
		end
		c_ts = ildb.get_ts(c_ts_id)
		c_rseg = route[i]
		c_lckp = {}
		
		if c_ts.route then
			if not try then atwarn("Encountered ts lock during a real run of routesetting routine, at ts=",c_ts_id,"while setting route",rtename,"of",signal) end
			return false, "Section '"..c_ts.name.."' already has route set from "..sigd_to_string(c_ts.route.origin)..":\n"..c_ts.route.rsn, c_ts_id, nil
		end
		if c_ts.trains and #c_ts.trains>0 then
			if not try then atwarn("Encountered ts occupied during a real run of routesetting routine, at ts=",c_ts_id,"while setting route",rtename,"of",signal) end
			return false, "Section '"..c_ts.name.."' is occupied!", c_ts_id, nil
		end
		
		for pts, state in pairs(c_rseg.locks) do
			local confl = ilrs.has_route_lock(pts, state)
			
			local pos = minetest.string_to_pos(pts)
			if advtrains.is_passive(pos) then
				local cstate = advtrains.getstate(pos)
				if cstate ~= state then
					local confl = ilrs.has_route_lock(pts)
					if confl then
						if not try then atwarn("Encountered route lock while a real run of routesetting routine, at position",pts,"while setting route",rtename,"of",signal) end
						return false, "Lock conflict at "..pts..", Held locked by:\n"..confl, nil, pts
					elseif not try then
						advtrains.setstate(pos, state)
					end
				end
				if not try then
					ilrs.add_route_lock(pts, c_ts_id, "Route '"..rtename.."' from signal '"..signalname.."'", signal)
					c_lckp[#c_lckp+1] = pts
				end
			else
				if not try then atwarn("Encountered route lock misconfiguration (no passive component) while a real run of routesetting routine, at position",pts,"while setting route",rtename,"of",signal) end
				return false, "No passive component at "..pts..". Please reconfigure route!"
			end
		end
		-- reserve ts and write locks
		if not try then
			local nvar = c_rseg.next
			if not route[i+1] then
				-- We shouldn't use the "next" value of the final route segment, because this can lead to accidental route-cancelling of already set routes from another signal.
				nvar = nil
			end
			c_ts.route = {
				origin = signal,
				entry = c_sigd,
				rsn = "Route '"..rtename.."' from signal '"..signalname.."', segment #"..i,
				first = first,
			}
			c_ts.route_post = {
				locks = c_lckp,
				next = nvar,
			}
			if c_tcbs.signal then
				c_tcbs.route_committed = true
				c_tcbs.aspect = route.aspect or advtrains.interlocking.GENERIC_FREE
				c_tcbs.route_origin = signal
				advtrains.interlocking.update_signal_aspect(c_tcbs)
			end
		end
		-- advance
		first = nil
		c_sigd = c_rseg.next
		i = i + 1
	end
	
	return true
end

-- Checks whether there is a route lock that prohibits setting the component
-- to the wanted state. returns string with reasons on conflict
function ilrs.has_route_lock(pts)
	-- look this up
	local e = ilrs.rte_locks[pts]
	if not e then return nil
	elseif #e==0 then
		ilrs.rte_locks[pts] = nil
		return nil
	end
	local txts = {}
	for _, ent in ipairs(e) do
		txts[#txts+1] = ent.rsn
	end
	return table.concat(txts, "\n")
end

-- adds route lock for position
function ilrs.add_route_lock(pts, ts, rsn, origin)
	ilrs.free_route_locks_indiv(pts, ts, true)
	local elm = {by=ts, rsn=rsn, origin=origin}
	if not ilrs.rte_locks[pts] then
		ilrs.rte_locks[pts] = { elm }
	else
		table.insert(ilrs.rte_locks[pts], elm)
	end
end

-- adds route lock for position
function ilrs.add_manual_route_lock(pts, rsn)
	local elm = {rsn=rsn}
	if not ilrs.rte_locks[pts] then
		ilrs.rte_locks[pts] = { elm }
	else
		table.insert(ilrs.rte_locks[pts], elm)
	end
end

-- frees route locking for all points (components) that were set by this ts
function ilrs.free_route_locks(ts, lcks, nocallbacks)
	for _,pts in pairs(lcks) do
		ilrs.free_route_locks_indiv(pts, ts, nocallbacks)
	end
end

function ilrs.free_route_locks_indiv(pts, ts, nocallbacks)
	local e = ilrs.rte_locks[pts]
	if not e then return nil
	elseif #e==0 then
		ilrs.rte_locks[pts] = nil
		return nil
	end
	local i = 1
	while i <= #e do
		if e[i].by == ts then
			--atdebug("free_route_locks_indiv",pts,"clearing entry",e[i].by,e[i].rsn)
			table.remove(e,i)
		else
			i = i + 1
		end
	end
	-- This must be delayed, because this code is executed in-between a train step
	-- TODO use luaautomation timers?
	if not nocallbacks then
		minetest.after(0, ilrs.update_waiting, "lck", pts)
		minetest.after(0.5, advtrains.set_fallback_state, minetest.string_to_pos(pts))
	end
end
-- frees all route locks, even manual ones set with the tool, at a specific position
function ilrs.remove_route_locks(pts, nocallbacks)
	ilrs.rte_locks[pts] = nil
	-- This must be delayed, because this code is executed in-between a train step
	-- TODO use luaautomation timers?
	if not nocallbacks then
		minetest.after(0, ilrs.update_waiting, "lck", pts)
	end
end


-- starting from the designated sigd, clears all subsequent route and route_post
-- information from the track sections.
-- note that this does not clear the routesetting status from the entry signal,
-- only from the ts's
function ilrs.cancel_route_from(sigd)
	-- we start at the tc designated by signal
	local c_sigd = sigd