Log for #openttd on 26th September 2021:
Times are UTC Toggle Colours
00:08:18  *** arikover has quit IRC
00:47:17  *** Progman has quit IRC
01:01:41  *** tokai has joined #openttd
01:01:41  *** ChanServ sets mode: +v tokai
01:08:21  *** tokai|noir has quit IRC
02:01:02  *** snail_UES_ has joined #openttd
02:18:07  *** snail_UES_ has quit IRC
02:28:58  *** WormnestAndroid has quit IRC
02:29:22  *** WormnestAndroid has joined #openttd
02:34:33  *** Tirili has quit IRC
02:39:06  *** Tirili has joined #openttd
03:02:35  *** debdog has joined #openttd
03:05:56  *** D-HUND has quit IRC
03:09:26  *** glx has quit IRC
04:19:44  *** _aD has quit IRC
04:39:18  *** Flygon has joined #openttd
04:47:59  *** Flygon has quit IRC
04:53:24  *** Flygon has joined #openttd
05:19:21  *** Tirili has quit IRC
06:17:43  *** andythenorth has joined #openttd
06:19:29  *** andythenorth has quit IRC
06:28:40  *** andythenorth has joined #openttd
06:56:17  *** EER has joined #openttd
07:03:55  *** nielsm has joined #openttd
07:16:33  *** Wolf01 has joined #openttd
07:23:11  <DorpsGek> [OpenTTD/OpenTTD] LordAro approved pull request #9575: Feature: Add use of Intel Intrinsics & RDTSC on e2k (MCST Elbrus 2000)
07:26:16  *** sla_ro|master has joined #openttd
07:46:25  <TrueBrain> wauw, explaining STUN is .. difficult
07:46:28  <TrueBrain> not sure it is worth it
07:46:48  <TrueBrain> (it is in our repo, but I was trying to make a news post)
07:47:54  *** gelignite has joined #openttd
07:48:08  <nielsm> "Instead of only B connecting to A, the game coordinator sets up to A connects to B and B to A at the same time, and this can trick the routers in-between to let the connection establish when it would usually be blocked."
07:48:20  <nielsm> I think that's reasonably correct and concise?
07:48:38  <TrueBrain> yeah, that part I have; the part where I went wrong is when I tried to explain why it sometimes fails :P
07:48:47  <TrueBrain> so I am just not going to :D
07:49:12  <nielsm> just leave that to the "this _can_ trick the routers in-between" phrasing
07:49:20  <nielsm> it can, it won't always
07:50:01  <andythenorth> isn't it just that thing where you have 2 phone calls on 2 phones, and you hold them up to each other?
07:50:10  <TrueBrain> kinda
07:50:17  <TrueBrain> but not really
07:50:18  <TrueBrain> :)
07:50:31  <andythenorth> that would be more like some proxy
07:50:32  <andythenorth> nvm
07:50:32  <nielsm> that's TURN, where you have a proxy in-between that relays
07:51:59  <nielsm> it's more like two rooms with 65535 hallways between them, each hallways has a door on each end and the doors are locked from the inside of the rooms
07:52:24  <nielsm> to meet up you need to agree on which hallway you open the door to from each room
07:52:59  <nielsm> and that's what the game coordinator does, it tells you which hallway you need to go into so you meet the other player
08:21:29  <DorpsGek> [OpenTTD/website] TrueBrain opened pull request #225: Add: news post about new Multiplayer Experience
08:26:08  *** Progman has joined #openttd
08:27:19  <LordAro> nice stuff
08:27:28  <LordAro> could do with a bit of copy editing, but looks good
08:28:11  <LordAro> not sure about your 20%/15% numbers though
08:28:28  <LordAro> isn't it 20% can't use direct ip, and 15% of those can't use STUN?
08:28:32  <LordAro> therefore...
08:28:36  <LordAro> @calc 0.2*0.15
08:28:36  <DorpsGek> LordAro: 0.03
08:33:17  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
08:33:43  <TrueBrain> tnx LordAro :)
08:34:03  <TrueBrain> @calc 100 / 62 * 13
08:34:03  <DorpsGek> TrueBrain: 20.96774193548387
08:34:12  <TrueBrain> 20.9% of the non-direct IP connections are TURN
08:35:20  <nielsm> line 93: "And the Game Coordinator can now coordinator "
08:35:27  <nielsm> coordinat_e_
08:35:30  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
08:35:52  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
08:35:56  <TrueBrain> fixed
08:36:31  <TrueBrain> LordAro: only 24% of the current connections are via Direct IP
08:37:00  <nielsm> 111: "Change a singleplayer game into a multiplayer game seamless" -- seamless_ly_
08:37:02  <TrueBrain> but I kinda assume that number will go down over time, as less servers will configure Direct IP :)
08:37:15  <TrueBrain> nielsm: please add it in the PR, bit easier to find for me etc :)
08:38:16  <TrueBrain> LordAro: <- these are the numbers btw; let me know if I can adjust my wording to make those more clear :D
08:40:23  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
08:49:35  <DorpsGek> [OpenTTD/website] James103 commented on pull request #225: Add: news post about new Multiplayer Experience
08:52:40  <LordAro> TrueBrain: i'd suggest that after release it will go up a bit, as most of the "proper" servers move over
08:52:46  <LordAro> but as you say, difficult to know
08:55:34  <DorpsGek> [OpenTTD/website] nielsmh commented on pull request #225: Add: news post about new Multiplayer Experience
08:55:41  <nielsm> here's a few suggestions
08:56:58  <DorpsGek> [OpenTTD/website] James103 commented on pull request #225: Add: news post about new Multiplayer Experience
08:57:24  <TrueBrain> tnx nielsm :)
09:03:04  <DorpsGek> [OpenTTD/website] TrueBrain commented on pull request #225: Add: news post about new Multiplayer Experience
09:03:36  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
09:04:20  <TrueBrain> did not apply all stylistic suggestions btw, hope you don't mind :)
09:05:32  <TrueBrain> but given the feedback is all in the details, I guess the main story is clear; good :)
09:11:41  <LordAro> :)
09:11:56  <TrueBrain> LordAro: <- "Add" or "Feature"?
09:12:32  <LordAro> add, imo
09:12:39  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain merged pull request #9575: Feature: Add use of Intel Intrinsics & RDTSC on e2k (MCST Elbrus 2000)
09:12:52  <TrueBrain> good, we are aligned :)
09:14:11  *** gelignite has quit IRC
09:17:17  *** Samu has joined #openttd
09:21:57  <Samu> hello
09:23:06  <Samu> I have a question regarding SmallStack
09:23:40  <Samu> since im increasing the StationID pool, do i also have to increase the 0xFFFD in this SmallStack accordingly?
09:25:05  <DorpsGek> [OpenTTD/website] LordAro commented on pull request #225: Add: news post about new Multiplayer Experience
09:26:42  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
09:26:48  <TrueBrain> funny how you changed things back how they were :P
09:27:45  <DorpsGek> [OpenTTD/website] TrueBrain commented on pull request #225: Add: news post about new Multiplayer Experience
09:29:13  <LordAro> :p
09:29:50  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
09:29:56  <DorpsGek> [OpenTTD/website] LordAro commented on pull request #225: Add: news post about new Multiplayer Experience
09:29:58  <TrueBrain> I trust your linguistic capability over mine ;)
09:30:55  <TrueBrain> "If our regular donations (and thank you so much for those!) run low (currently our donations look healthy), we will let you know, and run some kind of fundraiser." ?
09:32:35  <TrueBrain> no, going to rewrite it completely
09:34:17  <DorpsGek> [OpenTTD/website] TrueBrain updated pull request #225: Add: news post about new Multiplayer Experience
09:42:34  <TrueBrain> meh, switching to Cloudflare for is very difficult due to the complexity of the domain
09:42:49  <TrueBrain> kinda requires us to pay 200 dollar a month to use the more advanced setup method they have .. yeah, that is not realistic for us :D
09:43:23  <LordAro> wiki is not that expensive :p
09:58:11  <TrueBrain> meh, and I cannot test if it will work before changing the NS servers :P
09:58:24  <TrueBrain> guess what I can do is import the current zone file, and switch the NS servers
09:58:29  <TrueBrain> that should not change anything, basically
10:27:56  <TrueBrain> we really have too many subdomains :P
10:31:41  <nielsm> would it be simpler to move some of the things to separate second-level domains?
10:31:56  <nielsm> like and
10:32:38  <TrueBrain> it would, and we considered it when migrating to AWS
10:32:42  <TrueBrain> was ours for a while
10:32:48  <TrueBrain> the problem is that people trust
10:32:59  <TrueBrain> so in the end we voted against using other tlds
10:33:10  <TrueBrain> euh, not tlds, but you get what I mean
10:34:38  <TrueBrain> so I am now creating" target="_blank"> as domain where AWS fills it records .. and I am making a CNAME on Cloudflare from to" target="_blank">
10:34:48  <TrueBrain> it does mean people accessing services need to do 1 additional DNS lookup
10:34:58  <nielsm> not directly
10:35:09  <nielsm> usually the recursive nameserver clients use handle all that
10:35:32  <TrueBrain> who ever handles it, it is another roundtrip to another NS server :)
10:35:56  <TrueBrain> so if either Cloudflare or AWS DNS have a disruption, it will not work
10:35:56  <nielsm> yeah, but set the TTL reasonably and it gets cached
10:36:04  <nielsm> true
10:36:08  <TrueBrain> it isn't an issue on its own, just another layer
10:36:14  <TrueBrain> I was hoping I could avoid that, but .. I do not see how
10:36:28  <TrueBrain> so now I am manually entering ~150 CNAMEs in Cloudflare
10:36:51  <TrueBrain> and provisioning a shitton of domains on Route53 :P
10:37:24  <TrueBrain> I fully understand why Cloudflare wants to have full control of a domain, but it is annoying as fuck :P
10:37:42  <DorpsGek> [OpenTTD/website] michicc approved pull request #225: Add: news post about new Multiplayer Experience
10:38:26  <DorpsGek> [OpenTTD/website] TrueBrain merged pull request #225: Add: news post about new Multiplayer Experience
10:38:39  <TrueBrain> owh, lunch time! Almost forgot .. oops
10:51:55  <Samu> i have a question
10:52:09  <Samu> do I need to put parenthesis here? uint p1 = 1 << 8 | 1 << 16 | (GB(st, 16, 4) << 28);
11:01:48  <nielsm> IMO yes, even just for readability
11:02:03  <nielsm> separating things into visual groups
11:03:21  <nielsm> parentheses don't make anything slower or faster, but they affect readability. too many is tiring on the eyes, too few makes it require more effort to read the intended meaning
11:04:01  <nielsm> (okay, more parentheses might add a few microseconds to the compile time)
11:04:31  <TrueBrain> okay, that news post is a lot more text on the website than I expected, lol
11:04:54  <DorpsGek> [OpenTTD/website] TrueBrain created new tag: 1.4.36
11:09:42  <Samu> thx
11:21:09  *** sla_ro|master has quit IRC
11:30:03  *** virtualrandomnumber has joined #openttd
11:30:26  *** virtualrandomnumber has quit IRC
11:34:59  *** gelignite has joined #openttd
11:37:00  <TrueBrain> also published it on Steam, as there is where I noticed most questions about this :P
12:11:20  <TrueBrain> right, validating the DNS on Cloudflare is giving the same result as AWS ..
12:11:25  <TrueBrain> Python scripting ftw
12:13:13  *** glx has joined #openttd
12:13:13  *** ChanServ sets mode: +v glx
12:16:15  <glx> ShipAI, and its weird usage of buoys, triggered differences in paths returned by exitdir and trackdir, vs
12:17:55  <nielsm> glx: fyi I don't think I'll do any more tweaking on the drawpf code, and definitely not turn it into a PR
12:18:06  <glx> node count difference is huge
12:18:19  <TrueBrain> that is no surprise, ofc :)
12:18:20  <glx> yeah drawpf is a nice debug branch
12:18:46  <glx> but there's no way it could end up in master
12:19:53  <glx> I think I should just PR a commit revert
12:20:12  <glx> minor diffs in path, huge performance improve
12:21:39  <TrueBrain> go for it :)
12:22:18  <TrueBrain> and if trackdir is kept anyway with the node .. shouldn't a "tile only" also work?
12:22:33  *** Beer has joined #openttd
12:22:40  <TrueBrain> would drastically reduce it even further :P
12:23:38  <TrueBrain> (just setting the ExitDir to a fixed value might be a quick way to test that?)
12:25:07  <TrueBrain> next step: small penalty if you are near shores, to make ships not travel this annoyingly close to shores? I so hate that :P
12:29:00  <michi_cc> And reverse the penalty if the ship has prop 15 greater than prop 14 (canal and ocean speed fraction).
12:31:26  <Samu> do NewGRFs make use of StationID, Im iliterate regarding NewGRF code
12:31:51  <nielsm> no
12:33:00  <Samu> cool, that's a relief
12:33:13  <glx> not sure tile only would work, as I understand yapf (I may be wrong), each open node will lead to only one new node, with the lowest cost
12:33:23  <glx> so mostly only a straight line
12:34:24  *** tokai|noir has joined #openttd
12:34:24  *** ChanServ sets mode: +v tokai|noir
12:34:54  <glx> finally yapf is not that hard to understand, it's just split in so many files it's hard to follow
12:35:58  <nielsm> it's written as a framework to assemble pathfinders from
12:36:02  <nielsm> kind of
12:38:18  <TrueBrain> glx: I do not see why that would only lead to straight lines, I have to say :)
12:38:22  <TrueBrain> but maybe just worth a try
12:38:30  <Samu> are roadtypes 5 bits or 6 bits?
12:38:41  <Samu> i see some places with bits 5..9, others 5..10
12:38:44  <Samu> in comments
12:39:03  <TrueBrain> right, good thing I validated the DNS, found a few issues ... almost there :D
12:39:57  <glx> new tile is determined using old tile and old trackdir, you get 3 possible nodes on the new tile, with 3 different trackdirs, but using the tile only as key, you always discard 2 nodes
12:40:27  <TrueBrain> trackdirs = exitdirs?
12:40:46  <glx> 3 trackdirs can go to the same exitdir
12:41:14  <nielsm> exitdir is a tile edge
12:41:26  *** tokai has quit IRC
12:41:33  <nielsm> trackdir is both a track and direction along that track
12:41:40  <TrueBrain> let me rephrase: doesn't using only exitdir already dismiss that?
12:41:59  <glx> no because one entrydir goes to 3 exitdirs
12:42:20  <TrueBrain> either YAPF is a weird implementation of A* or I am missing something :P
12:42:31  <TrueBrain> if you are on tile A, you find neighbouring tiles, and give that a score
12:42:35  <TrueBrain> you put those on the open list
12:42:46  <Samu> how many roadtypes can there be? 31 or 63?
12:42:49  <TrueBrain> the lowest score survives if a node would only be a tile
12:42:54  <TrueBrain> that is the cheapest way to get on that tile
12:42:59  <nielsm> I think it might be better to think of vehicles going from tile edge to tile edge, rather than from tile to tile
12:43:31  <glx> on tile A, yapf uses the trackdir of the node to go to next tile (only one possible), then adds nodes that go outside of the new tiles
12:44:25  <glx> so each full tile has a minimum of 3 nodes, used to find more nodes
12:44:37  <TrueBrain> I have no clue what you just tried to say, sorry :D
12:44:57  <TrueBrain> "then adds nodes that go outside of the new tiles" <- I fail to parse this :(
12:45:25  <glx> when you enter the new tile, you can then follow 3 tracks
12:45:45  <glx> leading to the 3 tiles around the new tile
12:46:03  <glx> so 3 new nodes
12:47:03  <TrueBrain> okay, that I follow, and that is my point: I don't see the use of that
12:47:11  <TrueBrain> I should draw this out I guess :P
12:47:16  <TrueBrain> which is tricky
12:47:39  <TrueBrain> when A* is done properly, you don't need that much metadata to get the best route, mathematically
12:48:08  <glx> the base key for yapf is tile+trackdir
12:48:31  <TrueBrain> for trains, absolutely makes sense :)
12:48:40  <glx> tile+exitdir is just an optimisation, but that's how yapf works
12:48:43  <TrueBrain> for roads .. bit less so, but I assume it uses tile+exitdir already
12:48:50  <nielsm> the thing is that in the TTD movement model, the nodes are actually the center of tile edges, and the edges are tracks inside the tiles
12:49:27  <nielsm> but we don't have coordinates for the center of tile edges, we have coordinates for tiles, so the nodes are instead represented by the tile and the track+direction leading to that tile edge
12:49:50  <TrueBrain> that still means a tons of nodes are added for ship PF that are not relevant
12:49:58  <nielsm> probably yes
12:49:59  <TrueBrain> but I get what you say nielsm , tnx
12:50:18  <glx> yes and worse when using trackdir key for ships :)
12:50:30  <TrueBrain> what does roadpf use?
12:50:39  <glx> exitdir
12:50:41  <TrueBrain> pfew
12:50:52  <TrueBrain> but okay, basically YAPF abstracted train PF, and went from there
12:50:56  <glx> it's the only one that makes sense for road
12:51:08  <TrueBrain> as I said a few times already, I find YAPF very difficult to read because of the tons of abstraction on top of each other
12:51:18  <TrueBrain> it kinda loses the "I am used to A*, what are the f/g/h values" :P
12:51:43  <nielsm> if we end up making a new ship-specific pf can I suggest calling it Magellan?
12:52:03  <nielsm> (unless that name might carry too much negative baggage from the colonisation era)
12:52:10  <TrueBrain> :)
12:52:21  <TrueBrain> I have been looking into a mesh-based train PF
12:52:50  <TrueBrain> far less tricky than I expected
12:53:05  <TrueBrain> basically, create a single node out of a junction
12:53:13  <TrueBrain> and link those nodes
12:53:22  <TrueBrain> so track between junctions become irrelevant
12:53:46  *** andythenorth has quit IRC
12:53:56  <glx> lol just checked, for road yapf defaults to exitdir, but it's possible to use trackdir when disabling optimisations
12:54:04  <glx> just like ships
12:54:13  <TrueBrain> yeah .. "optimisation"
12:54:17  <TrueBrain> why can you disable that anyway? :P
12:54:19  <TrueBrain> PF has too many settings
12:54:22  <glx> except ships currently always disable optimisations
12:58:29  <TrueBrain> okay ..... I think I can request for the NS record of to be changed ... I did my best to validate everything ..
12:58:32  <TrueBrain> things might break :P
13:00:53  <TrueBrain> owh, Cloudflare support CNAME flattening
13:01:06  <TrueBrain> but only for the Pro plan, which is 20 dollar a month
13:01:21  <TrueBrain> that might be well worth the money, for the other goodies that gives
13:01:29  <TrueBrain> but for now .. lets first see what happens if we switch NS :)
13:05:58  <Samu> discrepancy in comment, bits 5..9
13:06:09  <Samu> bits 5..10
13:06:17  <Samu> which one is the right comment?
13:08:21  <nielsm> neither of them uses the roadtype value in p2 though
13:10:04  <Samu> it is passed through
13:10:31  <nielsm> I don't see the road type in either function, where do you see it?
13:10:59  <Samu> it is in p2 already
13:11:37  <Samu> sec
13:12:10  <Samu> it's done here
13:12:20  <nielsm> but either way
13:12:33  <Samu> ops 697 my ba
13:12:43  <nielsm> RoadType has 63 valid values, that makes it a 6 bit value
13:12:43  <Samu>
13:13:40  <nielsm> road_type.h line 30 you can see the MakeEnumPropsT final template argument is 6, that's the number of bits
13:14:32  <nielsm> EnumPropsT<RoadType>::num_bits is a constant with the value 6, from that declaration
13:14:48  <nielsm> if you allocate anything other than 6 bits for it it's a bug
13:15:27  <nielsm> so here's your homework, which of 5..9 or 5..10 is correct when 6 allocated bits is the only correct option?
13:15:53  <Samu> 5..10
13:16:24  <nielsm> yes
13:16:29  <Samu> _cur_roadtype << 5
13:16:37  <nielsm> that shifts it to position 5
13:17:25  <Samu> 63 then
13:17:32  <DorpsGek> [OpenTTD/OpenTTD] glx22 opened pull request #9576: Revert 7ca1793: Using Trackdir keyed node is not required, Exitdir keyed node still have the correct trackdir
13:17:51  <Samu> or 62, because 63 is for invalid road type
13:18:15  <Samu> but it starts counting from 0, so it's 63
13:19:51  <TrueBrain> I wonder if peter1138 remembers why he added it in the first place :D If he found scenarios that it broke etc
13:20:02  <glx> good question
13:20:05  <nielsm> there are 64 possible RoadType values, numbered 0 to 63, and the last value is reserved to mean "invalid"
13:20:06  *** jottyfan has joined #openttd
13:20:19  <nielsm> so 63 valid values numbered 0..62
13:20:28  <TrueBrain> but it is 18 months ago .. personally I wouldn't remember any of it :P
13:21:02  <glx> maybe it was just a misunderstanding of how ypaf works
13:21:11  <Samu> RoadType rt = Extract<RoadType, 5, 6>(p2);
13:21:29  <peter1138> There's a comment but I don't know what it means...
13:21:32  <TrueBrain> well, you are now our YAPF expert, so you have that going for you :)
13:21:45  <Samu> that reads 6 bits, so everything is right in code, no bugs
13:21:54  *** roadt_ has quit IRC
13:23:02  <glx> drawpf helped me a lot to see how things work
13:24:00  <glx> without visual hints it's way harder I think
13:24:18  <peter1138> IIRC it solved some issue, but I don't know what that issue was.
13:24:33  <TrueBrain> merge and find out again when someone runs into it?
13:24:57  *** jottyfan has quit IRC
13:27:48  <glx> <-- I'll need to make docking tiles visible to see if it's possible to solve this can't reach issue
13:28:06  <glx> maybe we need more docking tiles around docks
13:28:33  <peter1138> TrueBrain, also you missed a year, 30 months rather than 18 :-)
13:28:43  <TrueBrain> peter1138: oops
13:28:50  <peter1138> glx, it's easy to do, there is code specifically to limit it so that old behaviour was maintained.
13:28:55  <TrueBrain> 2020 didn't happen, right?
13:29:14  <peter1138> TrueBrain, pretty much!
13:31:35  <peter1138> glx, basically you can adjust IsValidDockingDirectionForDock() and put all DIAGDIR combos for both the two lines that are not zeros.
13:32:58  <TrueBrain> and now we wait for orudge to change NS servers .. pam pam pammmmmmmmmmm :D
13:33:08  <TrueBrain> I so hate changing these kind of things ..
13:33:48  <peter1138> Or change it 'return gfx >= GFX_DOCK_BASE_WATER_PART;' if we think it'd never be wanted to be set via NewGRF.
13:34:18  *** roadt has joined #openttd
13:39:37  <Samu> ais and human players can now build rail waypoints, rail stations, newgrf rail stations, road stops and docks!
13:39:43  <Samu> missing airports yet
13:40:24  <Samu> and then i have to find out if there are more obscure Commands that store StationIDs in p1/p2
13:44:08  *** roadt has quit IRC
13:44:20  *** roadt has joined #openttd
13:56:37  *** _aD has joined #openttd
14:06:20  *** WormnestAndroid has quit IRC
14:06:34  *** WormnestAndroid has joined #openttd
14:09:50  <Samu> ais can't chose airport layout
14:09:53  <Samu> interesting
14:10:21  <glx> peter1138: I can see many tests relying on GFX_DOCK_BASE_WATER_PART, so if newgrf come in play at some point IsValidDockingDirectionForDock() would not be the only impacted function
14:11:15  <nielsm> can we do something about the commands p1/p2 bitpacking some time soon?
14:11:53  <nielsm> at least wrap it in some way so you don't have to keep track of the packing everywhere, but better just change the protocol so there's more data available in general
14:12:10  <glx> like a big union ?
14:14:59  <nielsm> if p1/p2 is unavoidable for now, some kind of CommandData class wrapping all the data (including also tile and string) and with accessor functions
14:15:27  <nielsm> so you'd cmd.SetRoadType(x); cmd.GetRoadType();
14:15:30  <glx> oh no, I need to do savegame conversion for the new possible docking tiles
14:17:41  <TrueBrain> it wouldn't surprise me if Rubidium already had some ideas about this :P
14:19:32  <TrueBrain> okay, NS is changed for (tnx orudge ) .. now we wait to see what breaks :D
14:25:54  *** tokai has joined #openttd
14:25:55  *** ChanServ sets mode: +v tokai
14:27:05  *** michi_cc has quit IRC
14:27:55  *** michi_cc has joined #openttd
14:27:55  *** ChanServ sets mode: +v michi_cc
14:32:46  *** tokai|noir has quit IRC
14:35:33  <Samu> Order(packed) is here to ruin my day
14:37:04  <Samu> unless i make it int64 or uint64
14:37:21  <Samu> it's uint32 atm
14:39:43  <Samu> needs to pack more stuff
14:40:44  <Samu> CmdInsertOrder
14:40:53  <Samu> uses p2 for a packed Order
14:41:42  <Samu> it includes DestinationID, which is now increased in size
14:41:43  <glx> lol depots are valid docking tiles
14:41:56  <Samu> they can be
14:42:06  <Samu> if there's a dock
14:42:54  <glx> but there can be the depot wall blocking access to the dock
14:43:25  <nielsm> water tiles are a total mess
14:46:05  *** virtualrandomnumber has joined #openttd
14:46:05  <glx> ok just say there are doors in the wall so the ship can be loaded inside the depot
14:48:30  <glx> oh I know why not all tiles around the dock are allowed as docking tiles, it's because of the visual effect, there's a huge gap between dock and ship on dock sides
14:52:15  <TrueBrain> okay .. so Cloudflare is trying to be a pain in the ass .. they provision a certificate for you, which only works for a single subdomain
14:52:26  <TrueBrain> so would work, but to test it out, I of course try
14:52:32  <TrueBrain> they do not provision a valid certificate for that domain
14:52:36  <TrueBrain> wtf is this for nonsense ..
14:52:43  <TrueBrain> to fix it, we have to pay 10 euro a month
14:53:24  <TrueBrain> guess we should get an :P
14:53:25  <glx> weird, the certificate is not valid for all subdomains ?
14:53:34  <TrueBrain> not for subdomains of subdomains
14:54:09  <glx> silly
14:54:25  <TrueBrain> especially as they provision letsencrypt certificates
14:54:39  *** Strom has quit IRC
14:55:03  <michi_cc> Or it needs to be
14:55:21  <TrueBrain> yeah .. but that makes a lot of things a lot more annoying too :P
14:55:52  <TrueBrain> such silly limitations make things so much more difficult .. ugh
14:56:23  *** Strom has joined #openttd
15:01:59  *** Speeder has joined #openttd
15:03:35  <TrueBrain> meh, changing wiki.staging to wiki-staging is no fun on AWS .. as changing a certificate is not something that is made easy via CDK :)
15:07:45  *** Speeder__ has quit IRC
15:10:54  <TrueBrain> Hmm
15:11:32  <TrueBrain> if only I could take the lessons learnt from CDK and AWS, and do it all over again from scratch .... :P
15:11:37  *** michi_cc_ has joined #openttd
15:11:53  *** Beer has quit IRC
15:12:36  *** michi_cc has quit IRC
15:13:02  *** michi_cc_ is now known as michi_cc
15:21:53  <Samu> found another wrong comment
15:22:52  <Samu> says bits 24-31
15:23:06  <Samu> gets bits 20-27 instead
15:24:01  *** gelignite has quit IRC
15:26:39  <DorpsGek> [OpenTTD/OpenTTD] J0anJosep opened pull request #9577: Multi-tile depots
15:27:57  *** virtualrandomnumber has quit IRC
15:28:43  <glx> hmm yeah CommandData (and subclassing) with some kind of serialisation (can use the same format as now, but hidden) could be a nice thing
15:28:59  <glx> but it's a huge codechange :)
15:29:04  <nielsm> it'd be big
15:29:18  <nielsm> but it'd be for the better
15:29:23  <TrueBrain> ugh, I was like: lets email Cloudflare, see if they can help in any other way. Ticket is auto-closed with: DNS issue, see here. And that is it.
15:29:44  <TrueBrain> I understand they don't have that much time for people on Free plans, so it is not a complaint, but damn, this is not made easy :P
15:30:55  <DorpsGek> [OpenTTD/OpenTTD] glx22 opened pull request #9578: Change: Allow all tiles around docks to be docking tiles
15:33:49  *** Wormnest has joined #openttd
15:57:11  *** michi_cc has quit IRC
16:03:36  *** michi_cc has joined #openttd
16:03:36  *** ChanServ sets mode: +v michi_cc
16:03:55  *** WormnestAndroid has quit IRC
16:05:51  <Samu> Order new_order(p2 | ((uint64)GB(p1, 28, 4) << 32));
16:05:55  <Samu> gymnastics!
16:06:35  *** WormnestAndroid has joined #openttd
16:23:28  <michi_cc> Is there any reason (regarding commands send over the network) why the data needs to be tightly packed, i.e. is there some significant advantage for sending for example bools as a single bit (except a slightly higher needed data rate)?
16:24:43  <michi_cc> Serialisation command data only with byte granularity would probably to get away with a lot less code gymnastics.
16:24:58  <nielsm> I doubt it...
16:25:25  <nielsm> it's not like we're bandwidth limited by dialup modems any longer
16:25:44  <TrueBrain> but but but
16:25:46  <TrueBrain> 700 bytes/sec!
16:25:53  <michi_cc> Especially as the command rate is limited by human brains :)
16:26:05  *** andythenorth has joined #openttd
16:27:43  *** sla_ro|master has joined #openttd
16:27:48  <glx> scripts could fire many commands, but we limit them to be fair with humans
16:32:00  <Samu> tada! humans and ais may now insert orders to stations with index over 65535
16:32:19  <Samu> CMD_INSERT_ORDER exhausted all of p1 and p2
16:32:34  <Samu> lucky that there was still 4 bits free
16:32:51  *** blindvt has quit IRC
16:34:04  <peter1138> I started doing command data stuff but gave up quickly :p
16:34:23  <peter1138> And someone is bound to complain if the protocol changes ;-)
16:34:40  *** andythenorth_ has joined #openttd
16:34:44  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #9576: Revert 7ca1793: Using Trackdir keyed node is not required, Exitdir keyed node still have the correct trackdir
16:35:42  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #9578: Change: Allow all tiles around docks to be docking tiles
16:36:04  <TrueBrain> oof, savegame bump .. hmm .. can we backport that :P
16:36:16  <glx> I skiped the bump
16:36:29  <glx> forced conversion up to current version
16:36:44  <TrueBrain> is that wise?
16:37:00  <glx> minor extra loading times
16:37:14  <TrueBrain> why not bump the savegame, and we just backport that?
16:37:39  <TrueBrain> these kind of decisions tend to hunt us :P
16:38:10  <glx> well we are still in RC, so a bump may not hurt to much
16:38:27  <TrueBrain> the bump is more that the administration for it can be annoying
16:38:32  <TrueBrain> but we will survive ;)
16:38:44  *** Gustavo6046 has quit IRC
16:38:57  *** Gustavo6046 has joined #openttd
16:39:35  <michi_cc> Right now backporting a bump is still painless, it only gets 'tricky' if you want to backport a later bump without some other bump in-between.
16:40:02  <peter1138> 15:10 <+glx> peter1138: I can see many tests relying on GFX_DOCK_BASE_WATER_PART, so if newgrf come in play at some point  -- yeah, I have a patch for that ;-)
16:40:12  <glx> haha
16:40:16  *** andythenorth has quit IRC
16:41:00  <peter1138> and of course, the newgrf side just extends that array to be a property of dockspecs, so that particular function doesn't change much.
16:41:44  <DorpsGek> [OpenTTD/OpenTTD] glx22 merged pull request #9576: Revert 7ca1793: Using Trackdir keyed node is not required, Exitdir keyed node still have the correct trackdir
16:43:59  <peter1138> Ah #9578 removes it all. Shame.
16:44:26  <TrueBrain> if you have a better suggestion :)
16:44:57  <DorpsGek> [OpenTTD/OpenTTD] LC-Zorg opened issue #9579: Improperly assigned objects construction costs and HQ relocation
16:44:59  <glx> anyway, docking tiles already had a forced (and sometimes uneeded) conversion for some savegame version, it's just a new similar occurence
16:45:52  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain dismissed a review for pull request #9578: Change: Allow all tiles around docks to be docking tiles
16:46:09  <TrueBrain> I cannot believe I really thought it was a bump, I am really a shitty reviewer :P
16:46:28  <glx> peter1138: yeah it was simpler to remove the array for current state of code
16:47:02  <peter1138> More lines-of-code change though! ;)
16:47:51  <TrueBrain> Export Live-Staging-Wiki:ExportsOutputRefWikiService69FF224449441610 cannot be updated as it is in use by Live-Staging-Wiki-Policy and Live-Staging-WikiReload
16:47:56  <TrueBrain> I am so done with Cloudformation ..
16:48:07  <TrueBrain> any form of dependency is just ....... ugh
16:49:14  <glx> but I kept the now unused DiagDir parameter ;)
16:49:20  <peter1138> And removing it is not bad, if I ever get back to NewGRF docks that can reintroduce it.
16:49:53  <peter1138> Better to simplify and have less code, not leave stuff lying around in case it might be used in future.
16:51:51  *** lehu has joined #openttd
17:01:23  <DorpsGek> [OpenTTD/OpenTTD] glx22 updated pull request #9578: Change: Allow all tiles around docks to be docking tiles
17:02:29  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain commented on pull request #9578: Change: Allow all tiles around docks to be docking tiles
17:03:11  <glx> hmm there used to be a "bump" label
17:03:27  <TrueBrain> it wasn't used correctly, so I removed it :P
17:03:50  <TrueBrain> I removed many labels .. >:D
17:04:03  <glx> there are still too much :)
17:04:06  <TrueBrain> yup
17:04:18  <TrueBrain> and people tend to randomly put labels on stuff, without it actually helping anything :D
17:04:34  <TrueBrain> (I am one of those people :P)
17:04:39  <TrueBrain> I want "private labels" :D
17:07:35  *** lehu has quit IRC
17:08:39  <Samu> just finished looking at all commands, think my job is done!
17:09:12  <Samu> 32 files changed, just for increasing StationID pool
17:09:47  <Samu> and im not yet sure if I got all that needs change
17:11:14  <DorpsGek> [OpenTTD/OpenTTD] glx22 updated pull request #9578: Change: Allow all tiles around docks to be docking tiles
17:11:30  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #9578: Change: Allow all tiles around docks to be docking tiles
17:12:27  <glx> oh and indeed, we already did a bump between 1.11.0-RC1 and 1.11
17:12:51  <TrueBrain> yeah .. it happens :P
17:12:53  <Samu> oh, i've yet to look at afterload.cpp
17:13:01  <Samu> the non _sl.cpp file
17:17:33  <TrueBrain> owh, lol, our wiki explicitly doesn't cache normal pages
17:17:35  <TrueBrain> only images etc
17:17:36  <TrueBrain> but only for 5 minutes
17:17:41  <TrueBrain> guess that could use a bit of tuning too
17:18:25  <TrueBrain> <- otherwise seems to work fine (now cached by Cloudflare)
17:18:34  <TrueBrain> and no, you cannot login there, as that logs you in on
17:21:18  <glx> <-- doesn't seem to work ;)
17:23:35  <glx> ah and I just noticed alignment is off in "view source" mode
17:24:51  <TrueBrain> what doesn't seem to work?
17:25:19  <glx> "click me" does nothing
17:25:28  <Samu> wow this one was so simple
17:25:45  <TrueBrain> glx: yeah, no shit :P
17:25:53  <TrueBrain> if that worked, you should be worried :D
17:31:58  <DorpsGek> [OpenTTD/OpenTTD] glx22 merged pull request #9578: Change: Allow all tiles around docks to be docking tiles
17:33:09  <TrueBrain> I am so happy with AWS from time to time
17:33:23  <TrueBrain> yesterday, because of Hacker News etc, the load on the www spikes from ~1000 every 5 minutes
17:33:26  <TrueBrain> to over 10k
17:34:29  <TrueBrain> @calc 10000 / 5/ 60
17:34:29  <DorpsGek> TrueBrain: 33.333333333333336
17:34:34  <TrueBrain> lol
17:35:20  *** tokai|noir has joined #openttd
17:35:20  *** ChanServ sets mode: +v tokai|noir
17:38:57  <TrueBrain> right, lets just see what happens if we enable Cloudflare on .. it all seems to work as expected, so ..
17:38:58  <TrueBrain> YOLO!
17:40:01  <peter1138> You just lose IP address information, don't you? (Unless it's more than a basic account...)
17:40:33  <TrueBrain> how do you mean?
17:40:34  <Samu> check this out, it's huge
17:40:53  <TrueBrain> in access logs etc?
17:41:08  <peter1138> Yeah, all looks like it comes from CF.
17:41:24  <TrueBrain> yup; but we don't process access logs anyway :)
17:41:27  <peter1138> (Not necessarily a problem)
17:42:02  <TrueBrain> ideally I would like cloudflare to cache pages if the last-modified isn't changed, but I am not sure that is possible :D
17:42:16  *** tokai has quit IRC
17:47:05  <DorpsGek> [OpenTTD/OpenTTD] Castle4079King6023 opened issue #9580: [Bug]: Path signals are the only available signals to build
17:47:18  <glx> lol
17:49:38  <DorpsGek> [OpenTTD/OpenTTD] nielsmh commented on issue #9580: [Bug]: Path signals are the only available signals to build
17:49:41  <DorpsGek> [OpenTTD/OpenTTD] nielsmh closed issue #9580: [Bug]: Path signals are the only available signals to build
17:49:50  <TrueBrain> haha, I was doing the same thing nielsm , lol :D
17:49:58  <TrueBrain> my screenshot was more pretty, as it had paint-draw-red-circle on it :P
17:50:27  <Samu> that was a bit harsh
17:51:14  <DorpsGek> [OpenTTD/OpenTTD] nielsmh commented on issue #9580: [Bug]: Path signals are the only available signals to build
17:55:14  <peter1138> Better would've been to just completely remove non-path signals
17:55:38  <TrueBrain> you do that, let me find a bunker to shelter in :P
17:55:57  <TrueBrain> it is a sword some people are willing to die on :D
17:56:17  <glx> and they are vocal
17:56:57  <TrueBrain> right, images seem to cache as expected
17:57:03  <TrueBrain> the HTML files .. not so much
17:57:05  <TrueBrain> which is a bad sad :P
17:59:08  <TrueBrain> if I enable HTML files, it caches them for 2 hours, even if they change
17:59:11  <TrueBrain> that might not be ideal
18:01:50  *** Flygon has quit IRC
18:04:22  <peter1138> dpkg: warning: downgrading openttd from 12.0-beta2 to 12.0-RC1
18:04:33  <peter1138> Thanks dpkg :P
18:06:14  <TrueBrain> :D
18:06:22  <TrueBrain> has been an issue for as long as we have debian packages :P
18:06:36  <TrueBrain> hmm, okay, we send "cache-control: private, must-revalidate, max-age=0"
18:06:38  <LordAro> R < b
18:06:41  <TrueBrain> the "private" is a problem for Cloudflare
18:07:21  <TrueBrain> and we do this as you could be logged in or not
18:07:59  <nielsm> is "RC1" a downgrade from "beta2" because R has a lower ASCII value than b ?
18:11:35  <TrueBrain> right, lets see how much Cloudflare caches, and if that is an amount worth talking about
18:12:45  <andythenorth_> oh steam lol
18:12:57  <andythenorth_> not convinced TTD is better for single player than OpenTTD
18:13:04  <andythenorth_> but it's a big world, room for many opinions
18:13:36  <peter1138> heh
18:20:15  <TrueBrain> seems for I should hadd a cache-control header, but after that we can route it via Cloudflare too I think
18:20:23  <TrueBrain> but lets first give it a week or so for the wiki, see if it makes a difference
18:20:44  *** gelignite has joined #openttd
18:20:46  <TrueBrain> so far on the AWS side not much difference is seen, but it can still be old DNS etc
18:23:06  *** tokai has joined #openttd
18:23:06  *** ChanServ sets mode: +v tokai
18:28:34  <TrueBrain> 25 of the 65 languages are 100% done
18:28:49  <TrueBrain> my post in the Translators Team, it seems every reply pokes all the other translators again
18:28:57  <TrueBrain> that seems to work really well :P
18:29:19  <TrueBrain> also, I guess, it means several people left :P
18:30:06  *** tokai|noir has quit IRC
18:30:22  <TrueBrain> 20 languages have 0 translators
18:30:34  <LordAro> TrueBrain: self-maintaining lists!
18:30:38  <LordAro> even better
18:33:26  <TrueBrain> <- for who don't know what I am talking about
18:40:13  <TrueBrain> normally we push 50 MB per 5 minutes .. yesterday it reached 380 MB per 5 minutes :D
18:40:18  <TrueBrain> the slashdot effect .. its fun
18:40:40  <TrueBrain> CPU-wise, nothing happened
18:40:41  <TrueBrain> lol
18:41:51  <TrueBrain>
18:41:55  <TrueBrain> to put it in an image :P
18:42:45  <LordAro> nice
18:45:10  <LordAro> TrueBrain: also, i am helping.
18:45:27  <TrueBrain> helping with what? :)
18:46:38  <LordAro> translators
18:47:39  <DorpsGek> [OpenTTD/OpenTTD] DorpsGek pushed 1 commits to master
18:47:40  <DorpsGek>   - Update: Translations from eints (by translators)
18:47:49  <TrueBrain> how is that helping? :P
18:48:03  <TrueBrain> <- wow ..
18:48:34  <TrueBrain>  Showing with 771 additions and 106 deletions.
18:49:28  <Samu> who wants to test build more than 64000 stations?
18:49:44  <Samu> or rather 65535
18:50:07  <Samu> or review my code
18:50:13  <Samu> not sure how to ask
18:50:28  <TrueBrain> I think we already told you it has 0% chance of being accepted into master
18:50:49  <TrueBrain> so a review might not be the best use of our time :)
18:51:24  <Samu> I see :(
18:51:54  <Samu> was trying to figure out if I missed something
18:53:31  <andythenorth_> what is the limit in vanilla?
18:53:42  <Samu> 64000
18:53:58  <andythenorth_> and the increased limit is 65535?
18:54:19  <Samu> @calc 0xFF000
18:54:19  <DorpsGek> Samu: 1044480
18:54:23  <Samu> it's that
18:56:35  <FLHerne> 64k isn't an impossible number of stations to build on a 4096^2 map
18:56:57  <FLHerne> it's probably an unreasonable one though
18:57:10  <FLHerne> Are ship buoys 'stations'?
18:57:15  <Samu> yes
18:57:17  <FLHerne> I assume waypoints are
18:57:19  <Samu> they are
18:57:38  <Samu> that's what triggered me to increase this limit
18:57:40  *** J0anJosep has joined #openttd
18:57:52  <Samu> I couldn't max 5000 ships with 15 AIs
18:58:11  <FLHerne> ok, so you only need 16 columns of a waypoint every tile from one end of the map to the other
18:59:37  <Samu> if anyone dares to look/test/review, here it is
19:00:43  <TrueBrain> 20MB out of the 100MB of wiki data going to Cloudflare came from Cloudflare's cache .. that is not great :P
19:00:52  <TrueBrain> but I guess I should give it some more time :)
19:01:42  <TrueBrain> 50% of the served files were pngs
19:01:52  <TrueBrain> and 24% empty pages (I guess it means 304s)
19:01:53  <DorpsGek> [OpenTTD/OpenTTD] nielsmh started discussion #9581: Accessibility for motion-impaired users
19:04:59  <nielsm> does cloudflare know that the pages are entirely static content?
19:10:26  <andythenorth_> how cold are the caches?
19:10:39  <andythenorth_> does cloudflare have a discovery crawler, or does it wait for requests?
19:11:04  <andythenorth_> or is only seeing 20MB as cacheable?
19:11:19  <FLHerne> Samu: You should definitely add a helper function rather than copy-pasting those horrible expressions all over the place
19:11:38  <FLHerne> but looking at this, I agree there's no way this is being accepted
19:12:23  <TrueBrain> nielsm: it listens to the cache-control, so it is up to us
19:12:48  <TrueBrain> andythenorth: no, 20% came from cache, so possibly it needs to warm up more :)
19:13:08  <FLHerne> it adds a ton of complexity, uses up map bits, and doubles the memory usage of order lists for the sake of raising a limit that will never be hit in any conceivable human gameplay
19:15:05  <Samu> im running a 4k map, memory usage is about 1.3 GB, and it's running 14 AIs too
19:15:12  <TrueBrain> But I also think I should move logged in users to another URL or something.. allows caching HTML better :)
19:15:47  <andythenorth_> `/admin`
19:15:59  <andythenorth_> should I put a crawler over the site to warm up caches :P
19:16:03  <andythenorth_> maybe not
19:16:38  <TrueBrain> :p
19:17:37  <Samu> how do I check the memory usage difference?
19:18:24  <nielsm> with a spreadsheet
19:18:35  <nielsm> (don't measure, calculate)
19:19:09  *** Wormnest has quit IRC
19:19:52  <glx> with the recent ship pf change less buoys should be needed
19:20:41  <glx> and AI can learn to reuse buoys, ShipAI is silly with buoys
19:20:54  <Samu> but how did he figure it out so quickly that the memory usage would be doubled?
19:23:28  <J0anJosep> Hi! I know that PR9577 still has flaws and it is a really big PR, but I would like to have some opinions about functionality and coding.
19:23:36  <J0anJosep> I will change it from "draft" to "ready for review", although it really isn't. But I really need somebody to check it.
19:23:51  <glx> main issue I have is it's huge :)
19:23:56  <J0anJosep> Is it ok to change the "draft" state in this case?
19:24:34  <J0anJosep> +glx: I know. I could break it into smaller PR, but then the purpose of the changes is not clear
19:25:00  <glx> yeah sometimes it's not possible to split
19:25:51  <Samu> FLHerne, a helper function for the p1 and p2 parameters?
19:26:49  <Samu> something similar to Order::Pack()?
19:28:35  <FLHerne> Yes
19:28:35  <Samu> btw, I had to increase DestinationID from uint16 to uint32, but by doing so, inadvertedly had to increase Order:Pack from uint32 to uint64, though I only need 36 bits
19:29:21  <Samu> it's not doubled
19:30:12  <DorpsGek> [OpenTTD/OpenTTD] nielsmh commented on pull request #9577: Feature: Multi-tile depots
19:30:17  <J0anJosep> Well, if need be, I could make smaller PRs from #9577. Anyway, first I would need to know if there is any change of that feature being merged and then I could continue working on it.
19:31:55  <FLHerne> It is doubled
19:32:12  <FLHerne> even if most of the doubling is waste
19:32:32  <FLHerne> J0anJosep: I really hope there is, I love that feature
19:33:26  <FLHerne> nielsm: I believe your summary is wrong ;-)
19:34:00  <FLHerne> it does require that depots be long enough for their trains, unless I missed a fairly recent change
19:34:36  <nielsm> all the screenshots show single-tile depots holding length 4 trains at least
19:34:38  <FLHerne> magic one-tile depots excluded from that
19:34:58  <nielsm> well if the description text says that I missed it
19:35:15  <TrueBrain> Be careful this is a new PR :)
19:35:17  <FLHerne> oh, this is the wrong multi-tile depot PR
19:35:19  <FLHerne> sorry
19:36:00  <FLHerne> I meant "Feature: Extended depots", which is also by J0anJosep
19:36:08  <TrueBrain> I had to look twice too to figure that out :)
19:36:27  <glx> yes #8480 is #9574 + extra stuff :)
19:36:47  <glx> euh #9577
19:37:08  *** EER has quit IRC
19:39:13  <Samu> my dear 14x AIs have done it! They surpassed the 65535 barrier!
19:40:43  <DorpsGek> [OpenTTD/OpenTTD] J0anJosep commented on pull request #9577: Feature: Multi-tile depots
19:41:55  <FLHerne> still +1
19:42:16  <FLHerne> If people don't like the settings, just enable them both by default and hide/remove them, IMO
19:42:44  <FLHerne> "how can I depot-replace rail with maglev" is one of those perennial forum questions
19:43:07  <FLHerne> and anyone who doesn't want to can just...not
19:43:52  <glx> yeah it's a recurent question, conversion of a train fleet to new rail type
19:43:53  <FLHerne> I don't see how it gains any significant advantage in multiplayer
19:44:09  *** gelignite has quit IRC
19:44:21  <FLHerne> since rebuilding an existing network is economically stupid anyway
19:44:23  <J0anJosep> FLHerne: Yes, #9577 is different from #8480. In fact, #9577 is the first part of #8480. But splitting it, it will make things easier for everyone.
19:44:40  <FLHerne> J0anJosep: I understand now, just didn't realise initially
19:46:18  <glx> I guess "teleporting" requires manual action, not really pratical to abuse for human player (AIs will do it more easily)
19:46:56  <J0anJosep> #8480 had no chance in getting reviewed with 75 commits. So I have to split it. Also, splitting it, it will be easier to review and find bugs.
19:48:26  <FLHerne> glx: also, depot entry speed is really low
19:49:46  <glx> I didn't really read the code, but as it's stated you can't know where an aircraft will be when building on an airport with multiple depots, is it the same for trains ?
19:50:07  <J0anJosep> FLHerne: Speed can prevent the exploit in many cases, but even with it, moving wagons from one tile to another at no cost is an explotation.
19:50:52  <FLHerne> Yeah, it's whether the benefit is enough to make any difference
19:51:38  <FLHerne> I was imagining a chain of pairs of depots at maximum spread, where the train goes into one and leaves the other
19:52:16  <FLHerne> but you're crawling in and out at 38mph or something, and you'd need an AI to do it on any scale, and I think also to keep reversing the train replacement
19:52:31  <nielsm> like, I imagine sending a train into the depot at the bottom of a hill, and then pulling it out the depot 6 tiles up
19:52:35  <J0anJosep> +glx: It is the same for trains. But newly built trains are always built in the northern compatible depot tile. I am not sure if it is the northern compatible tile, or some other... but it is deterministic.
19:53:26  <glx> ok so it can be tricky and require extra steps before starting the train
19:53:34  <FLHerne> nielsm: sure, but you just stick like six engines on that train and it drives straight up the 6-tile slope at 150mph :p
19:53:46  <FLHerne> multiplayer servers hurt my eyeballs
19:56:49  <nielsm> FLHerne: I'm also thinking about early game where trains are weak and money is scarce
19:58:12  <glx> teleporting would be more effective with ships, they are slower than trains
19:59:28  *** WormnestAndroid has quit IRC
20:00:28  <J0anJosep> +glx: ships cannot teleport; road vehicles only teleport if replacement has a different road type and is not compatible with current tile.
20:00:43  *** WormnestAndroid has joined #openttd
20:01:09  <Samu> found issues, time to fix stuff
20:02:55  <andythenorth_> does the reviewing atomic commits thing really make it better?
20:03:17  <andythenorth_> TL;DR afaict all code review sucks, and isn't a great defence against future selves
20:04:51  <glx> reviewing commits one by one is simpler than the full stack of changes, but there's a possible downside: a later commit can change everything :)
20:05:46  <nielsm> still the state should make sense at any place
20:06:10  <nielsm> it's okay to have essentially dead code around that's intended to get used later, but not broken parts
20:06:13  <andythenorth_> genuine questions btw
20:06:27  <andythenorth_> I have read a lot about the pros and cons of TDD
20:06:38  <andythenorth_> but I've never found much interesting stuff to read about code review
20:07:13  <glx> anyway if a later commit rewrites a previous commit it's time to rebase things ;)
20:07:19  <nielsm> code review is perhaps more about readability and maintainability of the code than about correctness
20:10:04  <andythenorth_> I have seen a lot of them done in the last 15 years, but I couldn't say what makes a good one
20:10:07  <andythenorth_> only what makes a bad one
20:10:16  <andythenorth_> this is a live question in my life currently :P
20:11:01  <glx> hey sometimes we spot big mistakes (not always)
20:11:49  <andythenorth_> I don't just mean here
20:11:50  <andythenorth_> :)
20:11:55  <andythenorth_> things they tend towards:
20:12:02  <andythenorth_> - minor formatting changes
20:12:25  <andythenorth_> - nitpicks and matters of taste about formatting, implementation, and library or module or method choices
20:12:32  <andythenorth_> things they don't seem to prevent
20:12:35  <andythenorth_> - security holes
20:12:40  <andythenorth_> - bad UX
20:12:48  <andythenorth_> - race conditions in prod
20:13:10  <andythenorth_> things they don't seem to do as well they should
20:13:14  <andythenorth_> - peer learning
20:13:22  <andythenorth_> - asking if the design is good
20:13:35  <andythenorth_> - not accidentally making people rewrite things that have no real gain
20:13:49  <glx> bad UX is a usual thing with devs I guess, most hate doing UX stuff
20:13:58  <andythenorth_> most engineers can learn UX though
20:14:02  <andythenorth_> not all
20:14:04  <andythenorth_> but most
20:14:25  <andythenorth_> it's just integrity of concepts
20:14:46  <andythenorth_> are things well named, and is it clear what are primary entities, and what are secondary or tertiary
20:14:59  <andythenorth_> also UX is speed
20:15:10  <andythenorth_> people forgive some bad designs if it works fast
20:15:21  <andythenorth_> the rest is just flow
20:16:01  <DorpsGek> [OpenTTD/OpenTTD] simply-peachy opened issue #9582: [Crash]: Switching from fullscreen
20:19:35  <glx> oh that maybe because asserts are disabled
20:25:59  <DorpsGek> [OpenTTD/OpenTTD] glx22 commented on issue #9582: [Crash]: Switching from fullscreen
20:27:32  <glx> if I understand the trace in the crashlog I can only see one place in AddDirtyBlock() that could trigger a segfault
20:27:39  <nielsm> hmm random idea, add a list_paths console command that lists all search paths the game currently has for the various categories
20:27:57  <nielsm> it should probably only work on local console, not remote console
20:30:20  <LordAro> nielsm: sounds useful
20:34:34  <Samu> weird
20:34:46  <Samu> found a problem
20:35:16  <Samu> when I do this 1048575 << 12 I get 4 294 963 200, correct
20:35:48  <Samu> no wait, my bad
20:35:52  <Samu> gonna start again
20:36:55  <Samu> nevermind, visual studio is warning me about shift issues
20:40:15  *** J0anJosep has quit IRC
20:40:34  <glx> ok I tried a silly thing with multi-tile depot, I joint 2 depots on a different networks, not a good idea
20:41:18  <Samu> i figured it out, i have to (uint64)1048575 << 12
20:53:52  *** tokai|noir has joined #openttd
20:53:52  *** ChanServ sets mode: +v tokai|noir
20:57:26  <nielsm> made something:
20:57:46  *** sla_ro|master has quit IRC
21:00:50  *** tokai has quit IRC
21:03:08  <TrueBrain> Maybe also add something that lists the search path entries
21:03:35  <TrueBrain> And maybe call is lists_dirs or something.. paths can be a bit confusing with PF paths and patch etc
21:04:00  <nielsm> I went with paths to avoid the war of "dirs" versus "folders"
21:04:08  <TrueBrain> Lol
21:04:11  <Samu> had to cast them to (uin64)
21:04:25  <TrueBrain> We already have 'dir'
21:05:21  <Samu> the AIs were unable to insert orders above stationid 65535, stuff was being truncated
21:05:29  <Samu> time to rerun the AIs
21:06:47  <Samu> not sure if this->type also needs it, i guess not, visual studio didn't complain
21:15:28  <DorpsGek> [OpenTTD/OpenTTD] nielsmh opened pull request #9583: Add: Console command to list search directories for various things
21:18:49  <Samu> gotta sleep, cyas
21:20:52  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain commented on pull request #9583: Add: Console command to list search directories for various things
21:26:23  <DorpsGek> [OpenTTD/OpenTTD] nielsmh commented on pull request #9583: Add: Console command to list search directories for various things
21:26:56  *** Samu has quit IRC
21:27:16  <TrueBrain> Did not expect that to be an enum :p
21:29:02  <DorpsGek> [OpenTTD/OpenTTD] nielsmh updated pull request #9583: Add: Console command to list search directories for various things
21:31:18  <TrueBrain> Guess I should complain sentences end with a dot, but *shrug* :p
21:32:30  <DorpsGek> [OpenTTD/OpenTTD] TrueBrain approved pull request #9583: Add: Console command to list search directories for various things
21:39:49  <DorpsGek> [OpenTTD/OpenTTD] nielsmh merged pull request #9583: Add: Console command to list search directories for various things
22:07:02  *** andythenorth_ has quit IRC
22:09:51  *** nielsm has quit IRC
22:37:36  *** Wolf01 has quit IRC
22:45:53  *** Progman has quit IRC
23:06:31  *** Etua has joined #openttd
23:25:54  *** Etua has quit IRC

Powered by YARRSTE version: svn-trunk