IRC logs for #openttd on OFTC at 2020-04-21
⏴ go to previous day
00:04:49 <Xaroth> pypy is fun, but do keep in mind that most people who work with python don't even have it installed
00:05:38 <andythenorth> I only installed it for nml
00:07:42 <Xaroth> It might also be worth... cleaning up some of the codebase
00:08:52 <LordAro> i was wondering the same
00:09:07 <LordAro> it's not entirely clear why stringio is used at all, instead of just using an actual file object
00:09:39 <Xaroth> reduce complexity by offloading some stuff
00:10:30 <Xaroth> not sure what the progress thing is used for, but perhaps offloading the custom logging thing to the built-in python logging module
00:10:55 <DorpsGek> andythenorth: 0.166666666667
00:11:37 <andythenorth> on my relatively slow system
00:26:47 *** k-man__ is now known as k-man
00:43:56 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
00:59:26 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
01:06:25 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
01:52:59 *** Superleaf1995 has joined #openttd
03:01:19 *** tokai|noir has joined #openttd
03:01:19 *** ChanServ sets mode: +v tokai|noir
03:05:42 *** snail_UES_ has joined #openttd
06:12:01 *** Flygon_ has joined #openttd
06:13:10 *** k-man_ is now known as k-man
07:16:42 <DorpsGek_III> [OpenTTD/OpenTTD] DerDakon commented on pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/JfT6W
07:24:36 *** andythenorth has joined #openttd
07:49:23 <Barfoo> does anyone know how I can find out what bnc service is used to get chugchamp.crabdance.com in a user's whois?
07:49:30 <Barfoo> andythenorth any chance you know?
07:58:17 *** sla_ro|master has joined #openttd
09:15:04 *** andythenorth has joined #openttd
09:26:19 *** gelignite has joined #openttd
09:55:57 *** gnu_jj_ has joined #openttd
09:57:27 *** andythenorth is now known as Guest22939
09:59:17 <blathijs> LordAro: THanks for picking up nml again, I'm seeing lot's of activity :-)
10:16:39 <orudge> TrueBrain: Just wondering, are we on course for the termination of the OVH VPSes? Or do we need to migrate them somewhere temporarily?
10:22:37 <nielsm> yesterday it sounded to me like there were still a couple blockers that might prevent the in-game content download from working
10:25:50 *** arikover has joined #openttd
11:00:02 *** WormnestAndroid has quit IRC
11:00:15 *** WormnestAndroid has joined #openttd
11:07:33 *** arikover has joined #openttd
11:11:27 <TrueBrain> orudge : depends on the speed someone writes some CSS basically
11:11:40 <TrueBrain> Which is an odd blocker tbh, but here we are
11:12:18 <TrueBrain> So to be clear: the OVH server should be renewed, at least 3 months away. The two VPSes .. I will let you know this weekend :)
11:12:54 <TrueBrain> Their offer is pretty decent, so enough wigglingroom. But I hope we don't need it.
11:41:58 * Guest22939 reading "Do I need a CDN?"
11:42:06 *** Guest22939 is now known as andythenorth
11:46:54 *** rotterdxm has joined #openttd
11:53:52 <TrueBrain> Says a random guest :p
11:56:26 <LordAro> TrueBrain: no more weird than someone asking about a bnc service for some random znc instance
11:58:28 <TrueBrain> Yeah, that was very random. Happy he asked andythenorth about it :D
12:13:32 <rotterdxm> andythenorth, been using that FIRS test you sent me. the tropic climate doesn´t have any oil producing buildings anymore?
12:13:47 <rotterdxm> there is no producing industry in the ´display chain´
12:15:02 <andythenorth> I will fix, thanks
12:51:02 <peter1138> I'm sure I am hungrier when I have breakfast.
13:05:16 <Eddi|zuHause> i had lunch, anyway
13:05:36 <Eddi|zuHause> probably worth more corollas than you eat in a day :p
13:11:44 <andythenorth> I am the hungry caterpillar
13:12:11 <andythenorth> I have had 3 oranges, a bowl of cornflakes, half a packet of biscuits, vegemite on toast and a packet of hula hoops
13:12:18 <andythenorth> possibly I need lunch
13:13:41 <Eddi|zuHause> so when are you evolving into a beatiful butterfly?
13:13:52 <andythenorth> not in the last 40 years
13:17:17 <peter1138> I tried eating more but I just got fat :/
13:17:28 <peter1138> (That plus not cycling every day)
13:31:17 *** gelignite has joined #openttd
14:38:54 *** trainathon has joined #openttd
14:39:32 <trainathon> hi, is there a tutorial for drawing newgrf aircrafts? i can only found for trains and roadvechicles.
14:39:58 *** gelignite_ has joined #openttd
14:42:14 *** gelignite_ is now known as gelignite
14:49:16 <Eddi|zuHause> except for the wings, there's not much different
14:50:50 <trainathon> Eddi|zuHause: how can i than make the rounding of the aircraft? i struggle with that i my head.
14:51:17 <Eddi|zuHause> i don't think i can help you with that
14:52:33 <trainathon> I have yesterday ask to same qeustion on the irc for newgrf develepent and waited the hole day but no respons. i bin not yet begin with draw must i ask it there?
14:55:50 <Eddi|zuHause> look at some existing examples?
14:56:24 <trainathon> yes there bin i now for searching.
14:56:56 <trainathon> is it a good idea to scale down the photos for better scale?
15:05:57 <andythenorth> drawing planes is hard
15:07:18 <trainathon> i will just try it. i have done other stuff to begin with the hard stuff and following a tutorial
15:08:03 <trainathon> Whats a good scale per meter for a big aircraft? i have found that a big bus is 3.3 pixels per meter
15:08:25 <Eddi|zuHause> you will get nowhere if you try to figure out a scale like that
15:08:33 <Eddi|zuHause> the game scales every object differently
15:08:52 <Eddi|zuHause> you'll get gargantuan planes if you try to apply the scale of a bus
15:10:01 <trainathon> so i go for the trail and error method
15:10:30 <Eddi|zuHause> you should rather try to match the scale of the existing aircraft, either default or some other newgrf
15:11:17 *** snail_UES_ has joined #openttd
15:11:47 <trainathon> where can i look up the scale?
15:12:32 <Eddi|zuHause> by looking at the sprite and the size of the orginal plane that sprite is based on
15:12:51 <andythenorth> decompile it and look at the sprites? :)
15:13:34 <andythenorth> FWWI, I scale trains to around 32px = 60 feet
15:13:44 <andythenorth> which is near enough 3.3px per meter
15:14:12 <trainathon> andythenorth, i have donwloaded a newgrf with aircraft but if decompile the .tar file i get not working files
15:14:25 <andythenorth> you need to decompile the grf file with grfcodec
15:14:34 <andythenorth> then you get a spritesheet
15:15:36 <trainathon> is that also in th content+download folder?
15:16:05 <Eddi|zuHause> i was just playing a game and picked up a clipboard which says "gender-neutral polyamory-friendly partnership policy [unreadable fine print]"
15:20:25 <andythenorth> if you open the tar, there should be a grf file
15:21:43 <trainathon> so i need to unpakage the .tar file?
15:22:21 <trainathon> ok then i will do that
15:29:56 <trainathon> hooray for my i have succes full decompiled and extraxt the image. now i have a base to build on
15:54:32 *** andythenorth has joined #openttd
16:34:45 <Samu> but down there, there is no .find
17:06:08 <Samu> another win for ScriptPriorityQueue
17:07:28 <Samu> im comparing RailwAI's own priority queue, which is based on squirrel's own array
17:07:43 <Samu> with ScriptPriorityQueue oof michi_cc's
17:08:21 <Samu> looking at the dates, michi_cc's finds faster
17:08:33 <Samu> looking at system usage, it took also less cpu time, and less memory
17:10:25 <Samu> the reason I tested RailwAI's is that it's much more memory friendly compared to my memory hungry approach
17:10:56 <Samu> but it wasn't still enough to use less than michi_cc's
17:17:39 <Samu> just for comparison, mine takes 1m45s and uses 1.031.800K
17:18:05 <Samu> finishes search on 1st May
17:24:23 <Samu> I have a better screenshot, brb
17:25:30 <Samu> nielsm, your memory usage reports isn't that accurate
17:26:01 <Samu> it reports mine used 199 MiB, while in fact, system memory reports 1 GB
17:26:16 <Samu> but that's for the total of openttd, still
17:26:34 <Samu> easy to tell that it at least used 700 MB
17:41:38 <nielsm> those numbers aren't comparable
17:42:10 <nielsm> windows reports how much memory ottd has allocated from the system, that's the memory managed by the runtime library memory allocator
17:42:58 <nielsm> the script memory reports the total size of all current allocations performed by squirrel within one script environment, those allocations are taken from the runtime library memory allocator
17:43:10 <nielsm> but the runtime allocator also supplies memory to other things
17:43:23 <nielsm> and the runtime allocator typically never releases memory back to the OS
17:47:10 <trainathon> is there a way to see the location from a sprite that is in game where it is on the resource image (the original image where the sprite can from)?
17:47:54 <nielsm> no, the resource image no longer exists when the graphics have bene compiled to grf
17:48:09 <nielsm> the grf stores the pixel data for each sprite indididually
17:48:38 <trainathon> ok, than i must search by hand.
17:49:40 <nielsm> I'd like to have told you to use the sprite aligner to get the sprite index in a grf of a sprite you see on the screen, but it looks like that isn't possible
17:49:54 <nielsm> I don't know if there's any other tools for that
17:58:12 *** Wormnest has joined #openttd
18:10:34 *** Wormnest_ has joined #openttd
18:18:21 <nielsm> idea: can we make next version of ottd use a newer, more standard-like protocol to search and download content? and release it together with or shortly after the new bananas?
18:18:57 <nielsm> so the old content protocol will get less load over time, and the new one can be fully served by cdn or such
18:19:52 *** Lejving has joined #openttd
18:20:28 <andythenorth> nielsm it should be possible to correlate the sprite aligner and the spritesheet once the offset is known
18:20:32 <andythenorth> I never figured out how though
18:29:54 <trainathon> is there a hard limit on how big a sprite before inmporting it can by. The decompiled newgrf has spriteblocks of 88 pixels that is a lot bigger than for trains and busses of 32 pixels
18:30:22 <trainathon> not al the spites a 88 pixels width
18:31:36 <andythenorth> so large sprites will clip - showing through stuff they shouldn't
18:31:45 <andythenorth> with planes that rarely matters as they're flying
18:32:00 <andythenorth> but large sprites might clip through hangars or airport buildings
18:32:37 <andythenorth> but in AV8 / 9.8 pikka tends to use sizes that won't clip badly, so you can take it somewhat as a guide
18:32:37 <trainathon> so it is a good practies to cape with sprites of a max width of 32
18:32:48 <andythenorth> not for planes or ships no
18:33:09 <andythenorth> you can go larger, but you'll need to test in game
18:33:36 <trainathon> And yes the sprite was indeed clipping
18:34:03 <andythenorth> and they do clip ship depots, docks, and canal edges
18:34:08 <andythenorth> but big ships look cool
18:35:02 <andythenorth> also, note that the game compresses sprites in the / \ angles :D
18:35:18 <andythenorth> or rather, the game doesn't, but the historical art style does
18:36:00 <trainathon> than i will stick for now with the standard of 32 pixels than i can build my aircraft around 2 meter per pixel
18:38:00 <andythenorth> will that give you very small planes? :)
18:38:12 * andythenorth looks at original sprites
18:40:48 <trainathon> the original aircraft i will make is 92 meters long so i can do it to 1:1 ratio
18:45:37 <trainathon> is there a link were i can look to all the diffent orientation openttd need to work properly for the visuels?
18:49:11 <andythenorth> can't see an up to date one
18:49:29 <andythenorth> planes have extra takeoff/landing sprites I think
18:50:28 *** Progman has joined #openttd
18:52:47 <DorpsGek_III> [OpenTTD/OpenTTD] CheeseSprinkles commented on issue #8076: Odd Cursor Glitch When Moving Through Windows and Game https://git.io/JvhZm
18:53:30 *** supermop_Home has joined #openttd
18:58:21 <trainathon> andythernorth is there no extra sprites needed for navigation on airports?
18:59:43 <andythenorth> I can't find the docs for a definitive answer
19:00:31 <andythenorth> I'd suspect planes have 8 angles for 'travelling horizontal', and some more for take off / landing
19:02:20 <trainathon> then will a make 8 angles and 2 side for take-off / landing
19:06:19 <trainathon> i quit i have for today enough time spent on this.
19:07:17 <trainathon> bye thanks for all the good answers on my qeustion i will helpt my a lot further
19:31:27 <TrueBrain> "Smart" lockdown in The Netherlands extended for another month .. shocker :p
19:32:06 *** gelignite has joined #openttd
19:33:24 <luaduck> hm, we seem to be having a spate of people claiming that their company password has been broken
19:33:39 <luaduck> this has only started in the last week or two
19:37:49 <FLHerne> luaduck: 'broken' as in someone obtained it, or joined the company without it?
19:38:08 <FLHerne> luaduck: Or that it no longer works to access the company?
19:38:28 <luaduck> FLHerne: people joining passworded companies belonging to other players
19:39:22 <luaduck> user I'm working with right now claims they had a stupidly unique password set and yet another player joins the server, then 9 minutes later joins their company
19:40:55 *** frosch123 has joined #openttd
19:43:31 <FLHerne> And did something malicious?
19:44:43 <luaduck> (rules on our server dictate that the players comprising a company are responsible for all of the actions of that company, so password your shit)
19:45:38 <luaduck> slightly concerningly the player being accused has a username that doesn't resolve to much other than a security researcher in italy
19:45:51 <luaduck> that may be coincidental
19:47:14 <TrueBrain> does OpenTTD log when people try the wrong password?
19:48:46 <frosch123> didn't someone mention people getting kicked because of wrong company id?
19:49:13 <luaduck> I'll turn on debug logging at the start of the next game
19:49:23 * luaduck prepares more disk space in advance
19:49:27 <TrueBrain> how is the CPU and network traffic during those 9 minutes?
19:50:38 <TrueBrain> so yeah, logs surrounding these incidents would be helpful to say anything about it :)
19:50:49 <TrueBrain> it is a good signal btw; if we get more of these reports something is going on, ofc
19:51:33 <luaduck> yeah, more just reporting it as a possible
19:51:43 <luaduck> going to rotate all of our rcon passwords
19:52:29 <TrueBrain> that is always a good thing to do ;)
19:52:30 <FLHerne> frosch123: That's been happening quite frequently on Reddit S1 for a while
19:52:50 <TrueBrain> if you have any additional information, we would be more than happy to look into it
19:53:00 <luaduck> FLHerne: I believe that's being caused by our admin bot moving them out of the company a bit too quickly
19:53:37 <luaduck> before their client realises they're on team 255, so it tries to send a command for the company they think they joined and get kicked for abuse
19:53:55 <luaduck> TrueBrain: if I get more details I'll raise it on github
19:54:10 <TrueBrain> or if the logs should remain private, just give us a poke ;)
19:55:21 <FLHerne> luaduck: Ok, if so that seems not-relevant
19:55:42 <luaduck> yeah, that's been going on ages
19:56:51 <TrueBrain> I am going to spam this channel for a bit .. I would like to say: I am sorry, but I am not sure I am
19:57:26 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #21: Keep audit file per entry to see who changed what when https://git.io/JfkLj
19:58:23 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #22: Allow uploading .zip/.tar instead of individual files https://git.io/Jfktf
19:58:59 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #23: Use information from NewGRF / Baseset to prefill name/description/version https://git.io/JfktT
20:01:05 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #25: Read dependencies from NewGRF / Scenarios automatically https://git.io/JfktZ
20:01:49 <andythenorth> it's not very scientific
20:02:34 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #26: Allow uploading scenarios/heightmaps as replacement of an older https://git.io/Jfktn
20:03:42 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #27: Allow reading license/readme/changelog of packages via API (or give a link) https://git.io/Jfktl
20:04:15 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #28: Allow setting other authors during upload / edit https://git.io/JfktB
20:05:21 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #30: Allow selecting newer CC licences, and deprecate older https://git.io/Jfktu
20:05:42 <Xaroth> andythenorth: might be worth adding some debug/print lines to time specific sections of the code, so you can see a more distributed view of where time is spent.
20:06:04 <andythenorth> there is some basic timing approach in the code
20:06:17 <andythenorth> or there are 'proper' python timing modules I believe (never used them)
20:06:49 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #31: Allow setting versions to "new-games" / "savegames-only" if you like to https://git.io/Jfkt2
20:06:56 <andythenorth> but +1, I track performance in the code I write with simple print(time() - n) statement
20:07:47 <Xaroth> iirc pypy has also used some trickery for pickle to work around StringIO
20:07:54 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #32: API endpoints to get list of licenses and branches https://git.io/Jfktr
20:08:05 <Xaroth> which might be worth a shot as well
20:08:21 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #33: Check that \r\n are translated to \n properly https://git.io/JfktK
20:10:35 <andythenorth> is that a good or a bad thing :P
20:11:01 <DorpsGek_III> [OpenTTD/bananas-frontend-cli] TrueBrain updated pull request #1: Add: a very basic CLI frontend for BaNaNaS API https://git.io/JvjCk
20:11:32 <FLHerne> Xaroth: In this case, we're already timing one particular bit of the code
20:11:39 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 commented on issue #33: Check that textfiles have \r\n for user files and \n for ini files https://git.io/JfktK
20:11:41 <Xaroth> well, it's a tiny bit of trickery you can apply to just affect pypy, whilst keeping StringIO for 'normal' python
20:12:12 <FLHerne> The specific issue is that writing to StringIO is painfully slow in pypy
20:12:30 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 commented on issue #33: Check that textfiles have \r\n for user files and \n for ini files https://git.io/JfktK
20:12:33 <TrueBrain> frosch123: shouldn't we just convert all these files from \r\n to \n, like script-files, language-files, etc?
20:12:52 <frosch123> TrueBrain: i wondered what were the reason for the conversion
20:13:04 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain commented on issue #33: Check that textfiles have \r\n for user files and \n for ini files https://git.io/JfktK
20:13:14 <frosch123> only thing i came up with, that windows people should be able to read the readme with old notepad
20:13:25 <frosch123> that's why most coop makefiles did unix2dos on readme and stuff
20:13:26 *** gelignite has joined #openttd
20:13:27 <TrueBrain> pretty sure we converted \r\n to \n for pedenticness
20:13:47 <TrueBrain> euh, other way around
20:15:07 <Xaroth> FLHerne: correct, but for things like pickle they used a rather unobtrusive workaround, which is rather elegant
20:15:11 <TrueBrain> hmm, .txt to \r\n makes sense to support Notepad, indeed
20:15:26 <Xaroth> win10's notepad supports
20:15:34 <FLHerne> Xaroth: Yeah, I'm just reading up on that now
20:16:01 <FLHerne> That would work for us, I think
20:16:11 <TrueBrain> Win10 is not our minimum in supported Windows versions ;)
20:16:36 <DorpsGek_III> [OpenTTD/bananas-api] frosch123 approved pull request #17: Fix: use the same safe_name() routine as bananas-server https://git.io/JfkqT
20:17:03 <TrueBrain> frosch123: our ini-loaded supports \r\n fine, so I am not sure why the ini would be made \n, other than pedantic
20:17:14 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain merged pull request #17: Fix: use the same safe_name() routine as bananas-server https://git.io/JfT0F
20:17:34 <TrueBrain> guess to make it all the same? (me? you? EVERYBODY)
20:19:12 <TrueBrain> frosch123: musa already didn't do the conversion anymore
20:19:20 <TrueBrain> how about we remove it from the TOS and stop doing it?
20:23:51 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain commented on issue #33: Check that textfiles have \r\n for user files and \n for ini files https://git.io/JfktK
20:23:51 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain closed issue #33: Check that textfiles have \r\n for user files and \n for ini files https://git.io/JfktK
20:24:10 <luaduck> intriguingly we also have a desync issue, so more debug logging :V
20:34:23 <luaduck> dumb question, what logging should I expect to see on net debug level 2?
20:36:07 <frosch123> you want desync debug level 2
20:36:46 <frosch123> or 3 if you have enough disk space
20:37:06 <frosch123> not net debug level
20:40:11 <luaduck> I'm on net=2,desync=3
20:40:17 <luaduck> nothing interesting on stdout
20:41:02 <frosch123> look into autosave folder
20:41:29 <frosch123> desync writes directly to a commands-out.log
20:45:25 <luaduck> was more interested in the net output for catching the password thing
20:45:34 <luaduck> I do have a desync stored though, but I forgot to set an initial save
20:45:41 <TrueBrain> I am not sure that is logged
20:45:42 <luaduck> is the initial save / load absolutely essential?
20:45:45 <TrueBrain> we receive a partial copy/paste of some code
20:45:50 <TrueBrain> but I don't know what that was about :D
20:46:26 <frosch123> luaduck: with level 3, no
20:47:01 <luaduck> neat, I'll bundle commands-out.log and dmp_cmds* up and submit a github issue
20:53:57 <DorpsGek_III> [OpenTTD/OpenTTD] duckfullstop opened issue #8093: Regular server desynchronisations from clients https://git.io/JfkmV
21:03:54 <Xaroth> andythenorth: what do you use to test nmlc's functioning, besides `make test`?
21:11:38 <Samu> and that code prevents that "snake-biting-itself"
21:12:02 <andythenorth> Xaroth besides make test, I compile my own grfs
21:12:12 <andythenorth> it's not really comprehensive enough
21:12:22 <Xaroth> I'm... trying something.
21:12:31 <TrueBrain> I suggest you run while you can ;)
21:12:43 <andythenorth> there are grfs in /examples
21:13:36 <andythenorth> I don't actually know what the tests cover, it's unclear if compiling real grfs adds any coverage or not :)
21:26:19 <frosch123> Xaroth: i don't know what you are testing. but if it is performance, you should test with "nuts" and with "firs"
21:26:29 <frosch123> both are big grf, but very different
21:26:42 <frosch123> entirely different performance profiles
21:27:05 <frosch123> if you need a third tiebreaker, use xussrset
21:32:56 <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/fhbqc
21:42:48 <Xaroth> Well so far the normal regression tests seem to work
21:54:18 <Rayman> Hi there, just wondering if there is any help needed to translate OpenTTD? I've seen there are several strings missing in German. I wrote to translator@openttd.org one week ago to offer help, but didn't get an answer yet, so I'm asking here again. I'm also available in Discord, I wrote in #development channel.
21:54:57 <frosch123> the translator stuff will change in a few weeks
21:55:30 <frosch123> then applications will be handled better
21:56:19 <Rayman> OK, that's the point. Thank you.
21:57:23 <Rayman> If there is help needed after that, I'm still available to help
21:58:02 <frosch123> thanks :) there will be a news post when the translator is ready
22:06:32 <frosch123> what's the right thing to do? adjust the CSP, or remove the inline-css?
22:07:32 <frosch123> though possibly the CSP cannot be adjusted, and will always block inline-css :p
22:08:06 <frosch123> it is meant to block executing of javascript from foreign sites
22:08:10 <LordAro> inline css is a bit icky anyway
22:08:16 <frosch123> but apparently it is also used for css
22:08:41 <andythenorth> I don't like to comment on security because I know nothing
22:08:56 <andythenorth> but I would be surprised if style attr can be used for any interesting attacks
22:09:06 <andythenorth> that would seem remarkably tinfoil
22:10:02 <andythenorth> not being able to use any inline css is quite restrictive, but some people seem to like having their code split in 2 places, so eh :)
22:10:09 <andythenorth> call it a feature, move on?
22:12:56 * andythenorth looks what's in the css
22:16:07 <andythenorth> frosch123 it's nicely responsive :)
22:16:34 <frosch123> oh, i didn't know our https is verified by amazon, not letsencrypt
22:17:39 <LordAro> wait, that actually is a <marquee> element
22:17:46 <LordAro> i thought everyone removed support for those
22:18:30 <frosch123> they probably jit it to js
22:18:51 <LordAro> probably, but i thought it went the way of <blink>
22:19:04 <DorpsGek_III> [OpenTTD/OpenTTD] DerDakon commented on pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/Jfk3Q
22:20:19 <andythenorth> was the marquee the easter egg? :)
22:20:51 <frosch123> and yes, i added it one weekend, forgot about it, and scared myself the next weekend
22:21:12 <frosch123> btw. did anyone read my blabla on the frontpage?
22:22:26 <LordAro> frosch123: i know the capitalisation is there for a reason, but seeing "Newgrfs" & "Noais" does seem a little wrong
22:23:28 <frosch123> i copied that from the old website
22:23:52 <frosch123> no idea who ready frontpages
22:24:35 <frosch123> parts of the old frontpage were outddated, and lied :p
22:25:16 <DorpsGek_III> [OpenTTD/OpenTTD] DerDakon commented on pull request #7270: Introduce CMake (and removing all other project-related code) https://git.io/Jfksv
22:29:16 <michi_cc> When are we switching to C++17? I need a std::optional :( Somebody™ forget a way to distinguish between empty and unset std::strings.
22:29:54 <LordAro> just use a pointer :p
22:30:43 <michi_cc> std::unique_ptr<std::string>> looks really ugly, though.
22:31:26 <frosch123> use an alias then :p
22:31:57 <frosch123> template<class T> alias optional<T> = unique_ptr<T>; :)
22:32:22 <frosch123> it looked so weird :p
22:34:16 <LordAro> if you're going that method, template<class T> class optional<T> { bool valid; T data; }
22:35:29 <michi_cc> Well, if you want to get creative and assume OTTD string usage, the string "\0" could server as an undefined setinel :)
22:37:11 <nielsm> the solution is clearly to invent our own string class
22:37:41 <nielsm> a serious C++ project can't be without its own string class!
22:42:12 <frosch123> we had our own vector class
22:47:37 <andythenorth> oof flask projects build so easily :P
22:47:48 <andythenorth> unlike $other python web frameworks :P
22:52:31 * frosch123 wonders how andy is going to render the next newgrf using flask
22:53:19 <andythenorth> I quite like the flask templater
22:53:26 <andythenorth> but I'm used to chameleon
23:00:56 <Xaroth> py3.6's fstrings are also quite useful
23:01:11 <andythenorth> template languages are such a niche sport :)
23:01:38 <andythenorth> the one in Jekyll (Ruby) is quite interesting too
23:02:19 *** WormnestAndroid has quit IRC
23:02:32 *** WormnestAndroid has joined #openttd
23:06:00 <Xaroth> Make a small patch they said, it'll be easy, they said.
23:13:38 <FLHerne> Xaroth: What patch are you doing?
23:15:04 <Xaroth> andythenorth: ^ passes regression tests, if you want to throw your grfs at it, let me know if stuff breaks.
23:15:08 <frosch123> meh, f"" is a misfeature
23:15:38 * frosch123 likes to separate msgids and parameters
23:16:21 <Xaroth> frosch123: I found it weird getting used to it, but it made formatted strings a lot more readable once I got used to it
23:16:52 <Xaroth> because instead of having to match the param to the position in the message, you just see the param at the right position.
23:17:26 <FLHerne> Xaroth: I think you could drop the `:d` from most (all) of those
23:17:43 <andythenorth> Xaroth do I need to build something extra? "ModuleNotFoundError: No module named 'nml.util'"
23:18:16 <andythenorth> I just added an entire venv to a repo :P
23:18:17 <Xaroth> andythenorth: ammending said commit with said new file.
23:18:46 <andythenorth> I should not develop after 9pm, just watch TV :D
23:19:02 <Xaroth> mind, if you want to try it with pypy, make sure you run at least 7.2, as that supports 3.6
23:19:36 <Xaroth> also, there's many, many more things to replace >_>
23:19:47 <andythenorth> 7.3.0 so all good cheers :)
23:20:15 <Xaroth> 199 results for .format(), 17 more that use string concatenation.
23:21:00 * andythenorth does some timing runs
23:21:59 <LordAro> yeah, i've found :d to never be necessary
23:22:08 <LordAro> wasn't really necessary for .format() either
23:22:41 <Xaroth> .format is quite smart as it is
23:22:58 <Xaroth> fstrings use the same logic afaik, so that should all work out fine
23:23:05 <FLHerne> It's only relevant if you're printing float values as an integer or someting
23:23:44 <andythenorth> Xaroth interestingly the current branch is about 3s slower than nml master :)
23:23:47 <andythenorth> sorry to bear bad news :)
23:24:05 <FLHerne> Oh, it doesn't even work then
23:24:33 <LordAro> Xaroth: also, "{str(...)}" is very different
23:24:54 <LordAro> not sure where "very different" came from
23:25:06 <Xaroth> yeah, already nuked that one
23:25:11 <andythenorth> pypy3 is 3s slower for the entire nml run
23:25:30 <LordAro> andythenorth: over how many runs?
23:25:55 <LordAro> i'd expect 3s to be in the "random noise" range
23:26:07 <LordAro> what is it over, 5 minutes?
23:26:24 <Xaroth> andythenorth: can you run `python -c "from nml.util import StringIO; print(repr(StringIO))"` ?
23:26:45 <andythenorth> I'm just comparing python 3.8
23:27:16 <andythenorth> LordAro compile is usually 30s
23:27:26 <andythenorth> it's fairly consistent, but sometimes there is variation
23:27:46 <andythenorth> it tends to be mostly single threaded, and the I/O isn't much
23:27:52 <andythenorth> and I have a lot of spare cores
23:28:31 <andythenorth> this isn't stuff that should block us releasing btw :)
23:28:37 <andythenorth> it's micro-optimising :)
23:29:27 <LordAro> it's not, as far as i'm aware :p
23:29:48 <LordAro> #93 still needs merging, other than that it's ready to go
23:30:40 <andythenorth> I can merge #93 now, on the basis TB has approved it
23:30:45 <andythenorth> happy for me to do that?
23:30:59 <LordAro> if you're happy with it :p
23:31:58 <DorpsGek_III> [OpenTTD/nml] andythenorth commented on issue #42: Document how to release nmlc, including building a windows binary https://git.io/Je8vQ
23:33:05 <andythenorth> Xaroth FWIW python 3.8 is zero change in timings, which I assume is good
23:34:35 <Xaroth> bah, linux-tools' perf doesn't work on WSL
23:38:11 <andythenorth> Xaroth I ran more pypy3 runs, the 3s was just local noise. With or without patch, timings are ~identical
23:40:48 <Xaroth> heh, nml is a lot faster on my server than on WSL
23:41:07 <LordAro> WSL is a bit shit for various reasons
23:41:17 <LordAro> which is basically a VM
23:41:30 <Xaroth> mhm, but as long as it's well integrated
23:43:31 <Xaroth> andythenorth: on my server I'm getting a consistent 1.5s speedup to run through `make test`
23:44:30 <Xaroth> I'm testing through `perf` , might need a bit fiddling to get that enabled
23:44:49 <andythenorth> I'm just using crude 'time' currently
23:45:22 <Xaroth> yeah; time works, but I was interested in some more details :D
23:45:45 <Xaroth> about 2-5% fewer context switches between the branches
23:46:23 <andythenorth> with nml master, make test is 28-30s, mostly just below 29s
23:46:56 * andythenorth runs a few with the patch
23:46:57 <Xaroth> master ran runs between 11.88 and 11.98s (12000-ish task-clock msec).. new branch runs between 10.18 and 10.24s, (10300-ish task-clock msec)
23:48:06 <Xaroth> now that the CPU is warmed up it goes down to 9.6s
23:50:04 <andythenorth> with the patch I'm getting the same 28-30s
23:50:21 <andythenorth> I've just double checked, the built nmlc does match your branch
23:51:00 <andythenorth> might be architectural differences
23:51:12 <andythenorth> or something else got faster in my version of pypy3? :P
23:51:54 <Xaroth> pypy should be picking up on the StringIO improvement though
23:56:39 <Xaroth> pypy is a lot slower to run through make test, hrnf
23:57:54 <Xaroth> ... because it's not using StringBuilder
23:58:59 <FLHerne> Xaroth: Did you post a patch to use StringBuilder?
23:59:13 <Xaroth> it's there, but I see my flaw
23:59:45 <andythenorth> sleeping mistakes have been made recently :P
continue to next day ⏵
![https://i.imgur.com/dC6P1om.png[/img]](https://i.imgur.com/dC6P1om.png%5B/img%5D){kind=link}
![https://i.imgur.com/rLJQCRw.png[/img]](https://i.imgur.com/rLJQCRw.png%5B/img%5D){kind=link}
