IRC logs for #openttd on OFTC at 2023-07-15
âī go to previous day
00:11:36 *** Extrems has quit IRC (Ping timeout: 480 seconds)
00:13:26 *** Extrems has joined #openttd
00:32:10 <Eddi|zuHause> i hate when that happens... saying "it's fixed" without leaving a hint what was actually wrong
02:01:09 *** Wormnest has quit IRC (Quit: Leaving)
02:16:25 *** debdog has quit IRC (Ping timeout: 480 seconds)
02:26:30 <Eddi|zuHause> well... obviously...
02:30:21 *** D-HUND is now known as debdog
03:24:19 *** Artea has quit IRC (Ping timeout: 480 seconds)
04:16:45 *** tokai|noir has joined #openttd
04:16:45 *** ChanServ sets mode: +v tokai|noir
04:23:45 *** tokai has quit IRC (Ping timeout: 480 seconds)
05:15:37 *** HerzogDeXtEr has joined #openttd
07:10:46 <peter1138> Oops, I sat here too long.
07:10:49 <peter1138> And it's not raining yet.
08:27:14 <peter1138> Now it is. Wait long enough...
08:36:22 <andythenorth> especially after 7 mins, it covers search strategies
08:43:42 <andythenorth> need a smaller mouse ð
10:10:13 <truebrain> traffic to one single blog-post on our domain .. lol
10:12:24 <peter1138> The MacBook could easily handle that.
10:13:53 <truebrain> traffic was 8 times the normal during that spike ð
10:15:03 <truebrain> what might be more importantly, it also meant a spike in downloads
10:15:10 <truebrain> about 3 times more downloads during that period
10:16:51 <locosage> no noticeable spike in mp activity though
10:16:59 <locosage> and Spiff effect is also over
10:17:29 <truebrain> what is kinda funny to see, that most of those visits weren't done by a browser
10:17:36 <truebrain> either an app or a python library
10:17:52 <truebrain> shows a bit how most of those users visit the web ð
10:17:52 <locosage> robots, they're everywhere
10:19:09 <truebrain> it also shows some .. I think an app, directs all traffic via a single location
10:19:25 <truebrain> like, it is visible they were individual users, but they all have the same source address
10:20:24 <truebrain> so ~60% of the visits was by app, but after that Firefox won from Chrome .. also shows a bit what kind of people visit HN ð
10:20:43 <truebrain> as in a normal day, Chrome is far more popular than Firefox ð
10:22:10 * andythenorth looks up 'self-referential'
10:22:32 <andythenorth> "I wrote a blog post about infra to survive HN traffic, then posted it to HN"
10:24:25 <truebrain> and that is the lovely thing about this kind of infra .. Cloudflare handled it with ease ð
10:24:34 <truebrain> as for them, this amount of traffic is not worth mentioning ð
10:25:09 <truebrain> another random fun fact: the most CPU consuming process are the TURN servers
10:25:14 <truebrain> which is kinda surprising to me
10:25:47 <truebrain> all they do is receive a packet, validate it is an OpenTTD packet, and send it on the wire again ..
10:26:08 <truebrain> it might be that my APM is being CPU intense here .. time to remove that anyway, as it was only useful in the beginning ð
10:30:03 <truebrain> Application Performance Monitoring
10:30:16 <truebrain> it basically traces how long things took on a per-request level, sampled
10:30:30 <truebrain> so you can find out if it is your query that is creating these long response times
10:30:34 <truebrain> or just a function you fucked up ð
10:41:09 <truebrain> hmm, looking over what functions I traced, I doubt it is the reason for the CPU usage on the TURN service ð
10:41:37 <truebrain> ah .. but I also added it to the protocol handling .. that makes more sense
10:50:05 <peter1138> So a bit like the vehicle performance monitors in OpenTTD... they take up a non-insignificant amount of processing time, depending on platform.
10:51:28 <truebrain> and APM is very useful if you, like with game-coordinator, you collect that over a massive amounts of data
10:51:33 <truebrain> shows very clear lines
10:52:10 <truebrain> but where with OpenTTD the impact is in the 0.01% CPU or so, with Python it is ... a lot heavier ð
10:54:58 <truebrain> (I btw knew it had an impact on CPU, which is totally fine ofc; but as we aren't actually doing anything with the result anymore .. time to remove it ð )
10:57:51 <truebrain> I will get there with this CI PR! Will just take me a few more attempts it seems ð
11:00:34 <truebrain> from what I read, CodeQL no longer needs the dependencies installed, but I have yet to experiment with that ð So for now .. we duplicate some code ð
11:12:42 <truebrain> I will leave merging of that PR till Monday .. to not disturb all active TURN sessions ð
11:20:16 <peter1138> No haproxy-style safe reloading... tut!
11:20:33 <peter1138> Actually yeah, that is just reload config, not restarting a new version ð
11:29:30 <andythenorth> what shall we do today?
11:34:47 <peter1138> I had a little halloumi.
12:24:09 <pickpacket> peter1138: eat bicycle and go salading?
12:26:05 <pickpacket> So OpenTTD is *not* an abbreviation?
13:16:53 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
14:09:19 <Eddi|zuHause> about as much as "googling" is not a generic term for "searching"
14:36:52 *** gelignite has joined #openttd
15:23:36 *** Wormnest has joined #openttd
15:28:24 *** gelignite has quit IRC (Ping timeout: 480 seconds)
15:40:49 *** squirejames has joined #openttd
15:40:49 <squirejames> It's *suspiciously similar* to a game which did have an abbreviation of TTD, but, OpenTTD itself is not one
15:41:18 <squirejames> Because * checks notes * legal stuff
15:43:26 <locosage> first rule of OpenTTD: you do not talk about TTD
16:01:23 <Eddi|zuHause> that isn't actually a rule :)
16:04:24 <Wolf01> The rules are written in the topic, plus "don't piss off the devs or they'll add features like NRT"
16:05:04 <peter1138> We don't add features, we only remove them.
16:06:10 <belajalilija> you should remove all signals apart from two way block signals
16:15:05 <peter1138> We remove the things that don't work... so we should everything except path signals.
16:15:18 <Eddi|zuHause> what feature can we remove today?
16:21:26 <belajalilija> peter1138: you could do it like C:S where you cant load a save game without closing and reopening the game
16:24:56 <peter1138> I don't know what C:S is.
16:30:45 <belajalilija> if you mod the game it makes loading save games without first closing and reloading C:S impossible
16:31:17 <peter1138> Oh, well we used to do that ð
16:32:59 <peter1138> Back in the days when you had to put NewGRF config in openttd.cfg instead of storing it in the savegame.
16:37:10 <Eddi|zuHause> belajalilija: that's a limitation of the game engine, lots of games have that.
16:37:34 <Eddi|zuHause> belajalilija: it has to do with loading and unloading .dll files
16:39:02 <belajalilija> i dont play lots of game so it was something weird to me
16:42:37 <Eddi|zuHause> anyway, that doesn't apply to OpenTTD as we don't allow modding through .dll files
16:42:59 <Eddi|zuHause> we have a clear indirection layer between game code and mods
16:51:07 <peter1138> That reminds me of assetto corsa... there's a launcher that is used to select which track and cars are going to be in the game, and that's it, if you want to change anything you have to exit back to the launcher and start again. Of course it's set up so that it doesn't look like a separate launcher, just the game's main menu...
16:51:45 <peter1138> And then there's Live for Speed, where you can downloaded modded cars and switch tracks ... while staying connected to a mutliplayer server.
17:10:33 *** gelignite has joined #openttd
17:10:33 *** geli has quit IRC (Read error: Connection reset by peer)
17:21:05 <andythenorth> motivation issues
17:21:21 <andythenorth> I want the payoff of making something
17:21:27 <andythenorth> without the bother of making it
17:44:53 <andythenorth> I should probably have had lunch
17:52:15 <peter1138> I had salad, and then some...
18:04:31 <peter1138> truebrain: Actually the impact of the vehicle performance monitor is rather more than 0.01% CPU...
18:09:05 <peter1138> (Kinda forgotten about)
18:09:20 <peter1138> Old enough that I was using 1x interface ð
18:24:55 *** ChanServ sets mode: +v tokai
18:27:07 <andythenorth> goes it grf or something?
18:28:33 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
18:39:24 <kamnet> truebrain: LOL *extortion". Damn that's brave.
18:40:03 <DorpsGek> - Update: Translations from eints (by translators)
19:06:12 <peter1138> Nice, .net mvc rendering different HTML for the same source depending on whether it's a View or PartialView... (or more likely depending on the model state which will be different)
19:23:43 <peter1138> Hmm, NewGRFs changing industry cargo types dynamically only seems to happen on creation, in which case it should be fairly simple to handle that special case.
19:24:55 <talltyler> I wonder if it would break games that change industry newgrfs during a running game even faster
19:25:44 <peter1138> I don't think so, the cargos are set up when the industry is created, not when the saveload is loaded.
19:26:25 <peter1138> So if the update is broken enough that the slots don't match, then it would still be equally broken ð
19:27:01 <peter1138> Fortuately the 16-in/16-out stuff is down by cargo type rather than slot number, so it's already fairly resilient. Might not work but shouldn't crash.
19:27:12 <peter1138> At least, not this bit of it. Famous last words 'n all.
19:27:58 <peter1138> Unfortunately for industryspecs which are built to the old 2/3 system, they are done by slot number. So need to be careful.
19:28:04 <peter1138> And that's probably a majority tbh.
19:30:59 <peter1138> Okay, handling this single probably unused use-case means that #11133 isn't totally redundant, at least ð
19:31:43 <peter1138> Where it says "allow older grfs to skip slots", the comment now means "allow older grfs to add a slot with CT_INVALID"
19:32:54 <peter1138> Thus you end up with 2/3 slots in use if 1) the industry doesn't use the 16/16 feature, and 2) the industryspec uses the feature to change cargo slots dynamically on build (instead of just in the fixed spec).
19:35:18 <peter1138> case 0x89: return this->industry->produced[variable - 0x88].cargo;
19:35:31 <peter1138> I mean, I can see what that does, but it's not really necessary is it...
19:36:41 <_glx_> 08 B[2] Types of cargo produced (-1 if N/A)
19:37:37 <peter1138> > case 0x88: return this->industry->produced[0].cargo;
19:37:37 <peter1138> > case 0x89: return this->industry->produced[1].cargo;
19:37:37 <peter1138> Is probably clearer on what is happening...
19:38:03 <peter1138> And yes, these are the older accessors, there are newer ones that work based on cargo type rather than slot number.
19:39:20 <_glx_> it's even not directly listed in grf docs
19:39:38 <_glx_> so most likely not in NML
19:42:19 <_glx_> I don't even see how they could be used (the grf should already know what the industry produces)
19:43:05 <peter1138> Oh it's not just those two, there's also cargo waiting, rate and production/transported history. They will be useful.
19:43:59 <_glx_> but 6F and 70 are recent
19:44:42 <peter1138> Yes, those are the sensible ones that go by cargo type.
19:44:55 <_glx_> while 88 and 89 don't have equivalent for 16in/16out
19:46:08 <peter1138> Okay, but I'm not really sure what you're getting at? These are old variables from TTDPatch that are supported, and I just think the way it uses "variable - 0x88" to get 0 and 1 is a bit weird ð
19:46:43 <_glx_> yeah, it's how we used to do it
19:47:18 <_glx_> but I agree, 0 and 1 is simpler
19:48:15 <_glx_> the variable - 0xXX happens a lot in newgrf stuff
19:48:21 <peter1138> Anyway, vector .at() will throw an exception, so slightly safe than [], but not really going to do us much good ð
19:48:53 <peter1138> Unless it's something other than "variable - ", then actually no, it doesn't happen a lot.
19:49:15 <_glx_> I remember seeing it often
19:51:00 <_glx_> the 5 switches don't count ð
19:51:02 <peter1138> Just industries and stations. The - 0x80 in newgrf_engine.cpp doesn't count, that's a different reason.
19:52:09 <_glx_> looks like only cargo stuff uses this construct
19:52:30 <peter1138> station goods one is the same thing
19:52:59 <peter1138> On the other hand, there are a lot more of those. ...
19:54:11 <peter1138> 8C to EC, with 8 vars, so thats the old 12 cargo types ð
19:54:44 <peter1138> It's kinda different I guess.
20:01:35 <peter1138> "Source model: Open-source[disputed]"
20:02:08 <andythenorth> flat docks maybe? ð
20:02:40 <_glx_> oh the python stuff is the same with mingw (they also use pacman)
20:03:18 <peter1138> JGR gets some... interesting support questions.
20:03:43 <peter1138> "This feature that's only in JGRPP doesn't work in vanilla" ... "..."
20:04:11 *** virtualrandomnumber has joined #openttd
20:04:45 *** virtualrandomnumber has quit IRC (Remote host closed the connection)
20:37:06 <andythenorth> goes it extend GS?
20:46:39 *** F_Quin has quit IRC (Quit: Page closed)
20:46:54 *** qwebirc238 has joined #openttd
20:47:03 *** qwebirc238 has quit IRC (Remote host closed the connection)
21:02:57 <andythenorth> maybe I should give up on GS
21:03:03 <andythenorth> grf is more fun ð
21:03:54 <peter1138> You could extend NewGRF.
21:03:59 <peter1138> Or extend company shares.
21:04:04 <peter1138> Or extend penis extensions.
21:19:10 *** nielsm has quit IRC (Ping timeout: 480 seconds)
21:19:27 <F_Quin> people actually use IRC nowadays?
21:21:38 *** gelignite has quit IRC (Quit: Stay safe!)
21:27:50 <_glx_> most of us switched to discord
21:33:09 <pickpacket> I also use discord, but I much prefer IRC if possible. A lot faster and leaner
21:33:47 <pickpacket> that said I really wouldn't enjoy IRC as much if I didn't use a thelounge.chat instance
21:34:48 *** olionkey has joined #openttd
21:34:48 <olionkey> no matter how new something is there is always going to be that one guy who is using the old system
21:34:55 <olionkey> Think there is an XKCD comic aboutit
21:41:03 <_glx_> of course editing python via github is not a good idea
21:42:37 <F_Quin> I would hang but I think its only devs here as well as long time community members
21:42:54 <F_Quin> I'm not on the fourums so I think I'll go
21:43:00 *** F_Quin has quit IRC (Quit: Page closed)
21:45:23 <_glx_> weird, black passed while flake8 failed for line length
21:46:31 <_glx_> oh ok black workflow doesn't work
21:49:00 *** keikoz has quit IRC (Ping timeout: 480 seconds)
21:49:33 <_glx_> hmm no it works, but for some reason black seem to ignore -l (even locally)
21:59:22 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
22:08:43 * pickpacket is relatively new and would love some other new people to hang around
22:10:33 <frosch123> also noone is on the forums :p
22:10:43 <truebrain> you asked for this _glx_ ð
22:11:07 <truebrain> _glx_: `-l` does work, but not all lines are done. PEP-8 allows lines to be longer, but flake8 is like: NO
22:11:17 <truebrain> happens with strings, for example. Black doesn't know hot to split them
22:12:21 <pickpacket> frosch123: I am ðĪŠ
22:15:31 <truebrain> tnx for fixing this _glx_ ð
22:16:45 <_glx_> it's nice, on push I get a remainder about the dependabot PR
22:17:37 <truebrain> it is a bit odd it didn't make a PR for it .. it did for some other repos ..
22:17:45 <truebrain> but okay, in 2 weeks we update everything at once \o/
22:18:28 <_glx_> well opened 20 hours ago
22:19:05 <truebrain> the security advisory; but not a PR to update it ð
22:19:14 <truebrain> pressing a button does so, but some other repos it did it automatically .. guess org-settings
22:19:41 <truebrain> it is a CVE that doesn't apply to us anyway
22:22:05 <_glx_> and the 2 existing dependabot PR are the same
22:22:32 <truebrain> yeah; will be fixed next month ð
22:26:16 *** DorpsGek has joined #openttd
22:26:16 *** ChanServ sets mode: +o DorpsGek
22:42:12 <peter1138> So the issue there is my WSL Debian install is... eating space and not freeing it.
22:44:16 <debdog> or, if that has not much effect, apt-get clean
22:45:31 <peter1138> So WSL uses a virtual disk image which automatically grows. As such, it doesn't automatically shrink just because files are removed.
22:47:42 <peter1138> I guess I make branches out of all my stashes, then push to github, then delete and rebuild it...
22:50:07 <_glx_> seems you can compact it with diskpart
22:57:05 <peter1138> How do I fixed newgrf_debug_data.h:277?
22:58:05 <peter1138> It's a list that references the industry produced/accepted arrays... which in this branch are no longer always 16 entries...
22:58:48 <peter1138> NIProperty doesn't appear to be used anywhere else.
22:59:19 <peter1138> So it looks like a generic framework to do something which then only has one user ð
23:01:44 <peter1138> (I found a 10GB Business Central docker image which I'm no longer using, so that freed up a bit of space :D)
continue to next day âĩ
