IRC logs for #openttd on OFTC at 2020-04-20
⏴ go to previous day
00:20:55 *** Gustavo6046 has joined #openttd
00:30:22 *** andythenorth has joined #openttd
00:51:40 *** WormnestAndroid has quit IRC
00:52:00 *** WormnestAndroid has joined #openttd
01:08:17 *** tokai|noir has joined #openttd
01:08:17 *** ChanServ sets mode: +v tokai|noir
01:11:42 *** andythenorth has joined #openttd
01:19:55 *** WormnestAndroid has quit IRC
01:21:19 *** WormnestAndroid has joined #openttd
01:23:41 *** WormnestAndroid has joined #openttd
01:27:21 <mcbanhas_> andythenorth: I recall reading in some of the iron horse docymentation that the metro EMUs age cargo faster as to simulate overcrowding. How is this affect achieved exactly? And does it affect the whole train or just the passengers being carried in these specific units?
01:33:42 *** Wormnest has joined #openttd
01:40:36 <dwfreed> mcbanhas_: andy isn't online right now
01:40:52 <mcbanhas_> i noticed already :)
01:45:22 *** andythenorth has joined #openttd
01:51:23 <mcbanhas_> andythenorth: I recall reading in some of the iron horse documentation that the metro EMUs age cargo faster as to simulate overcrowding. How is this affect achieved exactly? And does it affect the whole train or just the passengers being carried in these specific units?
02:07:07 *** andythenorth has joined #openttd
02:25:45 *** andythenorth has joined #openttd
02:47:50 *** andythenorth has joined #openttd
02:48:38 *** Gustavo6046 has joined #openttd
03:01:11 *** ChanServ sets mode: +v tokai
03:06:06 *** andythenorth has joined #openttd
03:06:14 *** Eddi|zuHause2 has joined #openttd
03:40:38 *** andythenorth has joined #openttd
04:07:18 *** andythenorth has joined #openttd
04:30:02 *** andythenorth has joined #openttd
04:52:25 *** andythenorth has joined #openttd
05:11:38 *** andythenorth has joined #openttd
05:29:50 *** andythenorth has joined #openttd
05:52:12 *** andythenorth has joined #openttd
06:16:18 *** andythenorth has joined #openttd
06:51:27 *** andythenorth has joined #openttd
07:46:17 *** Eddi|zuHause2 is now known as Eddi|zuHause
07:55:47 *** sla_ro|master has joined #openttd
08:02:13 *** andythenorth has joined #openttd
08:24:32 *** andythenorth has joined #openttd
08:48:25 <LordAro> andythenorth: you were part/joining a lot overnight
08:54:28 <DorpsGek_III> [OpenTTD/bananas-frontend-web] andythenorth commented on pull request #7: Changes to the page header https://git.io/JfUb6
08:54:53 <andythenorth> my laptop is asleep overnight, it is not supposed to be doing this
08:55:09 <andythenorth> oh it's powernap
08:55:31 <andythenorth> it's a "do things while asleep" feature
09:13:45 <LordAro> my legs are a bit sunburnt today
09:14:00 <LordAro> i guess being in the full sun for ~7 hours will do that
09:47:13 *** k-man__ has joined #openttd
09:50:14 <Eddi|zuHause> the news said that sunburn risk is higher than usual, which is blamed either on clean air due to shutdown or ozone hole
09:51:00 *** gelignite has joined #openttd
09:51:38 <LordAro> ozone hole? not in europe, surely?
09:51:52 <LordAro> and the australia one was closing up, last i heard
09:52:34 <Eddi|zuHause> there's an ozone hole at the north pole, which is slightly less severe than the southern one
09:52:48 <Eddi|zuHause> and "closing up" is a measure of decades
10:50:53 *** WormnestAndroid has quit IRC
10:51:06 *** WormnestAndroid has joined #openttd
12:13:26 *** Smedles has joined #openttd
13:19:02 <Samu> i dont understand arrays, nor constructors yet... i can't pass an array on a constructor
13:19:49 <DorpsGek_III> [OpenTTD/OpenTTD] DonaldDuck313 opened issue #8092: Swedish translation issue in the performance rating window https://git.io/JfTUv
13:21:09 <Samu> why is that when I pass an array to the constructor, it then turns into an integer?
13:21:45 <nielsm> I kind of want to make an implementation of the OverflowSafeInt type that uses compiler intrinsics on gcc and clang, it makes for such better code... https://godbolt.org/z/X9wHB3
13:22:10 <nielsm> (but microsoft does not have anything equivalent, as far as I can tell)
13:23:27 <LordAro> nielsm: well you'd need to provide a fallback for other "bad" compilers anyway :p
13:23:52 <nielsm> actully I think there is an intrinsic that can be used for it... hmm
13:27:06 <nielsm> isn't it enough to support gcc, clang, and microsoft...
13:29:35 <Eddi|zuHause> remember when we used to support compilers like gcc 2.95.ancient?
13:30:05 <Eddi|zuHause> for that one user on MorphOS who didn't have a newer compiler?
13:30:18 <nielsm> I like the way it handled selecting between the min and max values
13:31:08 <nielsm> the way clang generates INT64_MIN from INT64_MAX is also cute
13:32:44 <Eddi|zuHause> isn't min like max+1?
13:33:18 <Eddi|zuHause> i'm not fluent in assembler, so i have difficulty to spot what you mean
13:33:43 <nielsm> and it uses the LEA instruction to add and store result into a different register
13:34:26 <nielsm> LEA is actually meant as an address calculation instruction (load effective address) but because of the complex addressing modes in x86 it can also function as a general purpose arithmetic instruction
13:34:58 <Eddi|zuHause> well, if you be running the code through static single assignment, and then optimize for register pressure
13:35:19 <nielsm> and the usual ADD instruction has for form ADD a,b, which means a = a + b, i.e. overwriting the first source operand
13:35:27 <Eddi|zuHause> grammar is on point today
13:37:52 <nielsm> but really, I'm also interested in making that code use intrinsics because it seems it's not quite complete, at least it seems there might be some underflow cases it doesn't handle
13:39:06 <nielsm> but then, all platforms need to use 100% equivalent code with identical results for this
13:40:22 <Eddi|zuHause> that sounds tricky to prove
13:41:10 <Eddi|zuHause> back when i studied, "prove the correctness" basically meant "can't use optimization"
13:41:43 <Eddi|zuHause> because it was effectively impossible to also prove the correctness of the optimizer
14:08:54 *** doug16k_ has joined #openttd
14:09:41 <doug16k_> I managed to freeze the game with train orders
14:10:15 <doug16k_> conditional jump to goto-depot if equal to 100 % followed by conditional jump to goto-depot if not equal to 100% load
14:10:47 <doug16k_> with both branches after the goto depot
14:10:51 <doug16k_> the depot it is already in
14:12:32 <doug16k_> ctrl-c from the terminal that launched it won't even kill it
14:12:50 <doug16k_> -9 killed it of course
14:14:22 <milek7> why 1.7.1? it is rather old
14:14:42 <doug16k_> ubuntu pkg mamager version :(
14:15:08 <doug16k_> I do have the source unpacked. perhaps I should compile it?
14:17:36 <doug16k_> wait, that source is old too. getting from git
14:20:19 <doug16k_> I have to help the build along by creating directories
14:25:32 *** berndj-blackout has joined #openttd
14:29:31 <doug16k_> can't reproduce with one little circle of track and a depot with a steam engine in it. maybe fixed
14:34:24 <TrueBrain> andythenorth : we forgot to celebrate the 2 year anniversary of migrating to GitHub!
14:41:49 <doug16k_> is it expected that a train station right against a producer of something (let's say coal mine) puts nothing in the station until the first time a train tries to get it, then suddenly it starts to gain the resource?
14:42:50 <doug16k_> so if your orders are to go to full load, the first visit to every station is guaranteed to have nothing, even if you build it a year ago against a resource
14:52:39 <FLHerne> doug16k_: Otherwise you'd have problems if it was near more than one industry, or near a town, and you only wanted one cargo
14:58:40 <doug16k_> ah, ya, then it would build up the unused resources probably OP to have it gathering up untold tons of stuff
14:58:50 <FLHerne> And if the station isn't served by vehicles taking a given cargo for a long time, it'll stop receiving it again
14:59:13 <doug16k_> (it would build up if it were how I expected it)
15:00:08 <FLHerne> For cargos that are being handled, the amount received is based on the station rating, which drops depending on the amount of waiting cargo
15:02:48 <FLHerne> So if you take much less cargo than the industry produces, there'll be a pileup of waiting cargo, but eventually the rating will drop until it reaches equilibrium
15:03:18 <doug16k_> so keeping them as close to zero as possible increases their output?
15:04:25 <doug16k_> I have had over 1 million oil in docks before. is there a limit?
15:05:43 <doug16k_> does it cause backpressure on the oil rigs served by the boats going to the docks?
15:06:34 *** bughunter has joined #openttd
15:06:38 <FLHerne> The unit counts for oil are multiplied by 1000; i.e. 1 million gallons of oil is equivalent to 1000 tons of most things
15:07:06 <FLHerne> No, only the amount waiting at the first station affects the rating
15:07:42 <doug16k_> so a fleet of boats filling docks can increase the docks to what limit? INT_MAX? :D
15:07:53 <bughunter> hey, i think i have found a bug in the toy land map(think not specific).
15:07:58 <FLHerne> I'm sure there's some implementation-defined limit with transfer orders, but it's higher than is reasonably achievable
15:08:28 <FLHerne> Someone plays toyland?! :o
15:09:34 <bughunter> @FLHerne, it was a small experiment and a the begin i found this bug.
15:12:22 <bughunter> i have made a dock and haven in a like and connect a pre-excistend canal with a man hand canal to the sea and purchase a schip in the lake and now the schip is stuck in the lake.
15:13:30 <FLHerne> bughunter: Ships can't go up slopes in rivers, you must build a lock
15:13:44 <bughunter> as you can see the boat will over the land next to the pre-excitent canal
15:13:58 <bughunter> and it will not came out of it
15:14:32 <bughunter> @FLHenrne i will try that
15:15:10 <nielsm> by the way the cargo amounts for oil and other liquids are internally stored in the same units as other cargo, the multiplication by 1000 only happens in display
15:15:41 <nielsm> you can see the actual counts of oil cargo units if you set it to be displayed in m3 (SI units)
15:16:02 <bughunter> FLHerne: thnaks now it works i have say nohing about a bug
15:16:12 <FLHerne> bughunter: Ok, great :-)
15:16:14 *** snail_UES_ has joined #openttd
15:21:36 <Eddi|zuHause> <TrueBrain> andythenorth : we forgot to celebrate the 2 year anniversary of migrating to GitHub! <- let's get together at a central location and celebrate with cake... oh wait..
15:54:38 *** supermop_Home has joined #openttd
16:16:34 *** WormnestAndroid has quit IRC
16:20:09 *** WormnestAndroid has joined #openttd
16:28:04 *** WormnestAndroid has quit IRC
16:29:48 *** WormnestAndroid has joined #openttd
16:48:54 <Samu> I just managed to implement the concept of "used tiles" or "conflict tiles" to graph.aystar, I'm becoming a pro at this, lol
16:50:04 <andythenorth> TrueBrain 2 years :o
16:50:28 <Samu> useful when stuff that is planned to be built doesn't let other stuff to build over or under it, talking about locks vs aqueducts
16:52:21 <Samu> or even bridges over bridges, which isn't possible
16:53:03 <Samu> turns out, it detects plenty of conflictuous tiles, which got me by surprise
16:53:39 <Samu> so I'm investigating if this is actually working as it should, or if it's conflicting in error
17:10:07 <Samu> eh, it's conflictuing too much, doesn't seem to be working as planned
17:11:25 *** Wormnest has joined #openttd
17:45:45 <blathijs> Now that OpenTTD 1.10 is out, any more plans for a nml release? Otherwise I'll have to backport a fix to Debian in the next two weeks to prevent having Openttd and OpenGFX removed for build failures on newer Python versions :-)
17:54:15 <LordAro> blathijs: why would that get OTTD removed?
17:54:25 <LordAro> OTTD doesn't depend on nml at all
17:54:32 <LordAro> but yes, someone should do that
17:54:48 <LordAro> i should get back to my release workflow PR
17:58:44 <kuba> hi, is there a shortcut so that when i click on a vehicle, it will bring up the group its part of in the company vehicle list?
17:58:51 *** kuba is now known as Guest22857
18:01:26 <blathijs> LordAro: It depends on OpenGFX, which does depend on nml.
18:02:05 <LordAro> it doesn't *need* to depend on OGFX
18:02:40 <LordAro> i'll see if i can do something with it this evening
18:14:26 *** Progman has joined #openttd
18:50:22 *** frosch123 has joined #openttd
19:22:19 <TrueBrain> LordAro: if there is anything I can help with, let me know
19:25:02 <LordAro> but it's not super important
20:03:12 <nielsm> means my callback is doing _something_
20:03:28 *** andythenorth has joined #openttd
20:06:14 <andythenorth> LordAro so have we volunteered ourselves for nml release?
20:11:15 * andythenorth looks in PRs first
20:13:34 <andythenorth> I tried to fix it, but the fix is rejected so eh :)
20:13:42 <TrueBrain> LordAro: there is a solution in that issue; so why not use that?
20:16:45 <TrueBrain> andythenorth: regarding #68, I normally am not a fan of these: you solved it one way, but there is a better way, go do that. But in this case that really is a trivial fix
20:16:47 <TrueBrain> so why not just fix it?
20:16:59 <TrueBrain> the fix is not rejected, just the way you approached it is
20:17:54 <TrueBrain> not sure why you are passive aggresive about it :)
20:20:51 *** supermop_Home has joined #openttd
20:23:48 <LordAro> TrueBrain: because it's not a solution - the produced binary would still not be manylinux2010 compatible
20:24:14 <LordAro> i'm not asking "how do i make this pass", i'm asking "how do i get a manylinux2010 binary with py38"
20:24:26 <LordAro> because centos6 is still supported, dammit :p
20:24:37 <LordAro> and clearly we have nml users still running it
20:24:38 <TrueBrain> what is the word .. pedantic? :)
20:24:49 <TrueBrain> can CentOS 6 run Python 3.8?
20:25:06 <TrueBrain> anyway, for them there won't be a whl .. you cannot fix the world
20:25:14 <TrueBrain> now there are no releases .. not sure which is worse :D
20:25:36 <LordAro> sure it can, i've built it myself for work purposes :)
20:25:44 <LordAro> (and indeed i built py37 for centos5)
20:26:26 <TrueBrain> so how about you accept the 2014 solution for now, and make an issue in the repo telling you should support 2010?
20:26:49 <TrueBrain> for the same reason we don't have linux-generic .. if I would have tried to solve that issue, we still wouldn't be anywhere close to a new BaNaNaS :P
20:27:18 <LordAro> as soon as i've finished writing a large comment to my colleague explaining why a file that's 50% comment is a bad idea
20:27:27 <LordAro> s/large comment/large message/
20:27:35 <TrueBrain> I was about to say :P
20:32:01 <nielsm> nice, made a working depth check for oil rig construction too, in newgrf
20:32:55 <nielsm> I'll probably have to prepare a PR for nml too :P
20:43:45 <TrueBrain> see if you can break it in any way :)
20:43:53 <TrueBrain> upload FIRS you say? Iron Horse? :D
20:44:20 <LordAro> TrueBrain: the massive 32bpp sets would probably be a good test
20:44:36 <Xaroth> You get both upvotes and downvotes for the <marquee />.
20:44:51 <DorpsGek_III> [OpenTTD/nml] andythenorth commented on pull request #68: Fix: close image files after use during palette check https://git.io/JfTWm
20:45:02 <TrueBrain> LordAro: internally, I uploaded every BaNaNaS file I could find :)
20:48:02 <Xaroth> andythenorth: need a hand with that PR?
20:49:54 <LordAro> TrueBrain: well yes, but not via "the interface"
20:50:04 <TrueBrain> not via the webinterface, that is true
20:50:10 <TrueBrain> but I did via the CLI interface :)
20:50:36 <TrueBrain> I am good in finding bugs :)
20:50:55 <TrueBrain> and the empty url and tag is not needed ..
20:53:54 <frosch123> Xaroth: the goal was to make a site so ugly, so that someone would make a better layout. anyway, i wrote the marquee in one week, and then scared myself the other week :p
20:54:29 <Samu> look at memory usage reported by windows
20:54:35 <TrueBrain> frosch123: owh, that was your easter egg? I liked the gimmick :)
20:54:58 <frosch123> i really forgot that i had added that :p
20:55:16 <TrueBrain> I hope someone styles it soon
20:55:24 <TrueBrain> but I am happy it works :)
20:55:34 <TrueBrain> lot of small bugs to fix, but .. we can do that :)
20:57:24 <TrueBrain> frosch123: if you have the time, could you look at the frontend-cli PR, and even better: give it a spin? I would like to merge that PR too, so others (please, let there be others) can also join in to help :D
20:57:27 <frosch123> LordAro: if you are testing, try files that were not repackaged by the old bananas
20:57:51 <frosch123> TrueBrain: i have the server pr on the other tab :)
20:58:16 <TrueBrain> owh, shit, yeah, we have to do thatone too :P
21:02:10 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain updated pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JvhXd
21:02:13 <TrueBrain> (minor fix and version bumping)
21:03:42 <nielsm> how many old versions of ottd have you tested it against?
21:03:51 <nielsm> just to ask some random questions
21:05:59 <TrueBrain> the server-part, I have only tested against 1.10
21:06:07 <TrueBrain> but we are mostly curious about the web frontend atm :)
21:06:37 <TrueBrain> to test the server, I need to fix the AWS infra first .. which is a though nut to crack :(
21:06:55 <nielsm> well, you have ten days to get it working as a full replacement, as far as I've understood?
21:07:05 <TrueBrain> in the ideal world, yes
21:07:16 <TrueBrain> so we need testers for the frontend now :)
21:07:16 <supermop_Home> can i get paid money to take Oil today?
21:07:22 <TrueBrain> find everything that is broken before release :D
21:17:50 <nielsm> so, what's the ip address of the replacement for content.openttd.org ?
21:20:46 <TrueBrain> somehow what I tell isn't coming across I notice, so let me try again :)
21:21:01 <TrueBrain> we currently replaced the webinterface (bananas-api and bananas-frontend-web) and they are deployed
21:21:06 <TrueBrain> this currently needs testing
21:21:17 <TrueBrain> we also have bananas-server, whcih is not deployed within AWS yet (well, it is, just not reachable)
21:21:30 <TrueBrain> content.openttd.org is the DNS used for bananas-server; ergo, that is not deployed yet :)
21:21:40 <nielsm> okay, I had understood it as "supposed to be working, but not in scope for testing yet"
21:22:20 <TrueBrain> it is seriously complicated, the BaNaNaS infra :)
21:22:24 <TrueBrain> (the old one was even more complicated)
21:23:31 <TrueBrain> the problem with AWS currently is btw, that it can only loadbalancing HTTP/HTTPS for you over IPv6
21:23:40 <TrueBrain> so I have to build my custom loadbalancer for our custom protocols
21:23:56 <TrueBrain> if you like, however, you can test the new server locally
21:24:16 <TrueBrain> you can run bananas-server just fine on a local machine, and you can test it out with any client :) (just point content.openttd.org to 127.0.0.1)
21:26:41 <DorpsGek_III> [OpenTTD/nml] Xaroth opened pull request #94: Change: Convert some code flows to context managers. (Supersedes #68) https://git.io/JfT8C
21:31:24 <LordAro> i think that's everything
21:32:41 <TrueBrain> stupid GH ate a comment of mine
21:35:00 <TrueBrain> you know how to add secrets btw?
21:35:10 <LordAro> i do, but i require a planetmaker
21:35:15 <TrueBrain> LordAro: do you know what the repository_dispatch does?
21:35:28 <LordAro> i just copied it from the DorpsGek repo
21:35:38 <TrueBrain> just remove the remote_dispatch block :P
21:35:43 <LordAro> given that's what you referenced in a previous comment
21:36:24 <TrueBrain> pretty sure I didn't mean that block :)
21:36:43 <TrueBrain> repository_dispatch is a "webhook" GitHub offers, where you can trigger GitHub Actions by doing an HTTP call to a certain endpoint
21:36:55 <TrueBrain> wish I use to redeploy stuff, for example
21:37:09 <TrueBrain> although I am really unsure why Dorpsgek has this
21:37:14 <TrueBrain> sounds like a copy/paste error there too :D
21:37:39 <TrueBrain> "website" uses it, so we can retrigger it after a new nightly
21:37:46 <TrueBrain> seems not useful for Dorpsgek (and nml)
21:37:53 <TrueBrain> so sorry if I said that before, as that is just plain wrong :D
21:38:15 <LordAro> TrueBrain: nah, you just said "make it like the other repos", so i copied the first one i found :p
21:38:25 <LordAro> i still like my original one better
21:38:34 <TrueBrain> I agree; and indeed :)
21:39:32 <LordAro> can be merged now, but won't work until pm creates an API token for me
21:39:58 <LordAro> (or more accurately, before someone with access to the info@ottdc.org email address verifies an email, then creates an API token)
21:40:46 <TrueBrain> I can look in that mailbox :P
21:42:17 <DorpsGek_III> [OpenTTD/nml] LordAro approved pull request #94: Change: Convert some code flows to context managers. (Supersedes #68) https://git.io/JfT4k
21:42:22 <TrueBrain> that is a minor detail there :)
21:42:54 <LordAro> TrueBrain: in theory i could add that email address, and switch it to primary
21:43:08 <TrueBrain> could be good, to have it on info@openttd
21:43:15 <TrueBrain> as it now moved to OpenTTD space
21:43:17 <DorpsGek_III> [OpenTTD/nml] LordAro merged pull request #94: Change: Convert some code flows to context managers. (Supersedes #68) https://git.io/JfT8C
21:43:26 <LordAro> the pypi account will always(?) be openttdcoop though
21:43:30 <LordAro> i'm not sure that can be changed
21:43:37 <TrueBrain> create an OpenTTD one, and move the package?
21:43:42 <TrueBrain> (no clue if that is possible)
21:43:47 <DorpsGek_III> [OpenTTD/nml] LordAro commented on pull request #68: Fix: close image files after use during palette check https://git.io/JfT4I
21:43:48 <DorpsGek_III> [OpenTTD/nml] LordAro closed pull request #68: Fix: close image files after use during palette check https://git.io/JeMeP
21:44:29 <TrueBrain> I will make an openttd account anyway, if there isn't one already, for the upcoming openttd libraries :)
21:48:59 <Xaroth> andythenorth: feel free to poke me if you have questions wrt my patch
21:49:28 <andythenorth> I'm going to test it :)
21:49:45 <LordAro> TrueBrain: looks like i can add additional "owners", so let me know when you've created the account
21:52:27 <DorpsGek_III> [OpenTTD/nml] andythenorth commented on pull request #94: Change: Convert some code flows to context managers. (Supersedes #68) https://git.io/JfT4r
21:53:10 <LordAro> TrueBrain: "User 'OpenTTD' does not have a verified primary email address and cannot be added as a Owner for project"
21:54:08 <TrueBrain> so that can take up to 5 minutes, I guess
21:54:42 * andythenorth looks at nml issues
22:00:36 <TrueBrain> LordAro: it is done
22:01:10 <TrueBrain> PyPi has no "teams" concept?
22:02:21 <DorpsGek_III> [OpenTTD/nml] andythenorth approved pull request #95: Fix #58: Loosen Pillow version constraint https://git.io/JfTBv
22:02:58 <LordAro> andythenorth: you didn't give me chance to edit the readme :(
22:02:59 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfTBU
22:03:26 <andythenorth> oof I'm an idiot
22:03:50 <andythenorth> less haste more speed :P
22:03:52 <LordAro> ah well, 3.4 is utterly ancient
22:03:57 <andythenorth> less children yakking in my ear
22:04:13 <LordAro> and 5.2 is fairly old
22:05:46 <andythenorth> LordAro I'll fix the readme
22:07:42 <LordAro> TrueBrain: alright, added the openttd account as an owner
22:07:51 <LordAro> you'll have to create an api token though
22:09:33 <TrueBrain> LordAro: added to the NML project on GitHub as Secret
22:10:45 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfTBl
22:10:47 <LordAro> andythenorth: more or less, except changelog is even less important
22:11:15 <LordAro> in fact, it doesn't even have one
22:11:37 <LordAro> oh wait, it's in docs
22:13:14 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfTB2
22:13:40 <andythenorth> LordAro we could move it
22:13:56 <andythenorth> it was a bit of coop thing to have changelog there, not repo root
22:14:00 <andythenorth> all my grfs are same :P
22:14:29 <LordAro> .devzone needs removing as well, while you're at it
22:14:56 <frosch123> if you move the changelog, check whether the Makefile bundles it or something
22:15:09 <andythenorth> yeah, I wondered if it's used for dist
22:15:17 <TrueBrain> frosch123: spaces in file name is totally okay in 2020
22:15:19 <LordAro> makefile could go as well, tbh
22:15:36 <TrueBrain> (I even have a space in my folder these days to spot errors caused by it .. :D)
22:16:02 <TrueBrain> hmm .. guess the name and version should be slugified too?
22:16:03 <frosch123> LordAro: makefile runs unittests
22:16:21 <andythenorth> LordAro there's quite a bit of stuff that looks bundling related
22:16:25 <andythenorth> I'm scared to touch it tbh
22:16:32 <andythenorth> and yes, makefile I use
22:16:39 <andythenorth> 'make install' is how I...install
22:16:58 <LordAro> and i guess a local install thing is kinda useful
22:17:48 <frosch123> TrueBrain: ouch, looks like openttd does not validate the filename, so you can put funny ".." in there
22:18:03 <TrueBrain> so I should sanitize it :)
22:18:15 <TrueBrain> what was the trick to do that .. hmm
22:18:25 <andythenorth> LordAro new plan, don't touch anything we don't understand yet :P
22:19:13 <TrueBrain> that moment you are sure you wrote a "slug" function, but cannot remember where and when
22:19:53 <LordAro> andythenorth: i can promise .devzone is not required :p
22:21:49 * andythenorth reads commit message rules
22:22:41 <DorpsGek_III> [OpenTTD/nml] andythenorth opened pull request #97: Cleanup: remove .devzone dir, no longer required https://git.io/JfTBj
22:27:16 * frosch123 found two bugs in the frontend :/
22:27:54 <TrueBrain> commit went through nicely
22:28:23 <TrueBrain> and I am trying to find a script I know should be there but I cannot find it :D
22:30:12 <frosch123> aw, i can no longer upload tars?
22:30:32 <frosch123> how do i upload a gs? all individual files?
22:31:14 <TrueBrain> is on the wishlist :)
22:31:30 <TrueBrain> and if someone who knows a thing or two about javascript is around, he can make it so you can drag all your files to the webpage at once
22:31:30 <frosch123> hmm, individual files does not work, since it drops folder structure
22:31:33 <TrueBrain> tus is made for that :)
22:31:48 <TrueBrain> so you cannot upload content with folder structure via the web atm :)
22:31:51 <TrueBrain> the cli does support it :D
22:32:14 <TrueBrain> I know for sure I wrote a function to make "safe" filenames
22:32:29 <frosch123> you could have written it again by now
22:32:40 <TrueBrain> but I didn't want to :P
22:33:57 <LordAro> wouldn't surprise me if there was something in the python stdlib
22:36:00 <TrueBrain> it is so annoying, I knew I wrote code for this, but I have no clue where :D
22:36:19 <TrueBrain> so who is to blame for that lapse in memory ..
22:37:15 <TrueBrain> there it is, in another repo
22:39:23 <TrueBrain> no; turns out I already figured thisone out in the API
22:39:27 <TrueBrain> as the foldername in the tar should also be the same
22:39:31 <TrueBrain> and that is based on the old code
22:39:33 <TrueBrain> so it stays the same
22:40:29 <frosch123> haha, i just noticed that the BaNaNaS repo will be at the top of ottd's github very often :)
22:40:44 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT0e
22:41:04 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain updated pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JvhXd
22:41:21 <TrueBrain> we can even make DorpsGek_III announce changes :P
22:41:41 <TrueBrain> the commitlog will look boring
22:41:58 <TrueBrain> - Update: changes made via content-api (by Librarian)
22:42:12 <TrueBrain> and I see I have to add the deployment keys via DorpsGek, otherwise I am named every time :P
22:45:02 <TrueBrain> there, fixed that issue at least :)
22:47:49 <frosch123> what is the commit interval?
22:47:58 <TrueBrain> for edits, 15 minutes
22:49:19 <TrueBrain> --commit-graceperiod SECONDS Graceperiod between commits to disk.
22:55:39 <TrueBrain> a "Revoke access" link might be nice, yes
22:55:47 <TrueBrain> I will turn the wishlist in issue btw tomorrow or so
22:58:09 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT0z
22:59:40 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT02
23:03:42 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain commented on pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT0K
23:06:50 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain updated pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JvhXd
23:07:33 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened pull request #17: Fix: use the same safe_name() routine as bananas-server https://git.io/JfT0F
23:07:57 <frosch123> ah well, let's just hope that noone has a name with only invalid chars
23:08:11 <TrueBrain> yeah .. that should fail validation :D
23:08:19 <TrueBrain> the foldername will be empty too
23:08:51 <TrueBrain> no, as I strip that :D
23:09:05 <TrueBrain> I will make an issue out of it :)
23:09:15 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 approved pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT0x
23:09:37 <frosch123> TrueBrain: it's no issue. otherwise you would have to check uniqueness
23:09:52 <TrueBrain> we can prevent those from being uploaded on the API
23:09:57 <TrueBrain> that was what I was aiming for at least
23:10:42 <TrueBrain> I am surprised there never has been a name collision btw
23:10:44 <TrueBrain> as that is pretty easy
23:11:03 <TrueBrain> okay, that is a bug :)
23:11:15 <frosch123> i made a change and reverted it within the commit period
23:11:36 <TrueBrain> I am honestly wondering, shouldn't we check filename are unique? :P
23:11:42 <TrueBrain> the client will rename the tarball to the name we give
23:12:09 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #18: Package with "name" and "version" of non-ASCII characters will have a filename "-.tar.gz" https://git.io/JfTEL
23:13:30 <DorpsGek_III> [OpenTTD/bananas-api] TrueBrain opened issue #20: GitHub commit message can have duplicated entries https://git.io/JfTE3
23:13:36 <frosch123> [23:10] <TrueBrain> I am surprised there never has been a name collision btw <- lies. there were two "slovenian town names", and before we renamed them in the database, clients got the filesize of one, and the content of the other, so it was always broken
23:14:01 <TrueBrain> and why didn't we fix it properly? :P
23:14:23 <frosch123> the developer closed the bug as "unreproducible"
23:14:41 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain dismissed a review for pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfT0x
23:14:41 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain updated pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JvhXd
23:15:08 *** Smedles has joined #openttd
23:15:28 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 approved pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JfTEC
23:16:01 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain merged pull request #1: Add: first version of the bananas_server rewritten in Python https://git.io/JvhXd
23:16:05 <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick commented on pull request #8091: Add: [Script] Native priority queue; useful e.g. for pathfinders. https://git.io/JfTE4
23:16:07 <TrueBrain> shit, forgot to add credentials
23:16:12 <TrueBrain> so it won't auto-deploy
23:16:15 <TrueBrain> owh well, that is for tomorrow
23:19:18 <TrueBrain> LordAro: remember those repository_dispatch? I copied it EVERYWHERE :D And I don't know why :P
23:19:49 <Samu> existing openttd is taking an hour
23:19:50 <TrueBrain> okay, I need to change something in this repo, to trigger a rebuild
23:20:03 <Samu> because of that stupid test of mine, seeing how much memory would it eat from my sys
23:20:21 <Samu> now my HDD is 100% usage
23:21:29 <frosch123> hmm, since dorpsgek is no longer owner, info@ does no longer get notifications about new keys
23:21:46 <frosch123> i guess everyone is happy about less spam :)
23:22:22 <TrueBrain> I cannot find a small thing to change in bananas-server, bah
23:24:16 <frosch123> run a spell checker over the docs?
23:25:51 <TrueBrain> meh; found another way to trigger the build
23:25:59 <TrueBrain> turns out having those repository_dispatch can be useful for something :P
23:26:44 <DorpsGek_III> [OpenTTD/nml] frosch123 commented on pull request #97: Cleanup: remove .devzone dir, no longer required https://git.io/JfTEQ
23:28:12 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain opened pull request #4: Fix: [Actions] deployment was not triggered for staging https://git.io/JfTEN
23:28:21 <TrueBrain> and the mandatory: shit, see, I did forget something :D
23:29:41 <DorpsGek_III> [OpenTTD/bananas-server] frosch123 approved pull request #4: Fix: [Actions] deployment was not triggered for staging https://git.io/JfTuJ
23:30:18 <DorpsGek_III> [OpenTTD/bananas-server] TrueBrain merged pull request #4: Fix: [Actions] deployment was not triggered for staging https://git.io/JfTEN
23:31:37 <_dp_> Samu, told you infinitely growing array can't be good :p
23:34:04 <FLHerne> andythenorth: Do we still want that patch for PyPy performance? (or an alternative one)
23:37:27 <andythenorth> StringIO is sloooow :)
23:38:45 <FLHerne> I think we probably need the same for grf output?
23:38:56 <FLHerne> At least, I can't see why it would be different
23:39:07 <FLHerne> In which case it probably needs factoring out somehow...
23:39:39 <FLHerne> The patched version is (marginally) slower than present with CPython
23:39:41 <andythenorth> I didn't test for grf
23:39:43 <Samu> technically, it's not infinite growth, but yeah...
23:39:53 <Samu> it's not cleaning itself
23:40:00 <FLHerne> By ~10% or something, IIRC
23:41:18 <DorpsGek_III> [OpenTTD/nml] andythenorth commented on issue #42: Document how to release nmlc, including building a windows binary https://git.io/Je8vQ
23:41:48 <andythenorth> FLHerne hmm, that's a regression for a minority case :|
23:41:53 <andythenorth> maybe we can't do that
23:42:08 <andythenorth> pypy is soo much faster for nmlc than py38 though
23:42:35 <FLHerne> I think it's probably worth it
23:43:03 <FLHerne> For any reasonably-sized grf, 10% of writing time is 0.1s or something
23:43:23 <FLHerne> And for unreasonably-sized grfs, pypy is the answer
23:43:23 <andythenorth> anyone who misses the 10% should really try to switch to pypy
23:43:38 <andythenorth> it's not hard these days, I just used the binary installer on mac
23:44:01 <FLHerne> On Arch, the various Python dependencies were slightly annoying
23:44:09 <FLHerne> Of course, I could just use pip...
23:48:12 <LordAro> i'm not convinced we should be working around bugs in an "alternative" "compiler"
23:48:36 <andythenorth> I'm not convinced either
23:48:47 <andythenorth> but it is a lot faster :P
23:49:17 <andythenorth> I can run it as a local patch only, but it gets complicated for testing nml :P
23:50:29 <andythenorth> if it's a flat NO I don't mind, but let's not waste FLHerne's time :)
23:51:00 <FLHerne> LordAro: I mean, the definition of cStringIO doesn't say how fast it's supposed to be
23:51:23 <FLHerne> LordAro: And if "not fast enough" is a bug, CPython is overall slower than PyPy even without this patch
23:51:34 <LordAro> the pypy PR linked referenced suggests that it is an issue
23:52:44 <andythenorth> I don't have benchmarks for the nml (non-nfo) case of this
23:52:49 <LordAro> how does your patch compare with andythenorth's "use an array" hack?, ooi?
23:53:09 <FLHerne> IIRC it's about twice as fast, and 10x faster than stringio
23:53:41 <LordAro> a nice detailed table is required, i think
23:53:43 <FLHerne> I don't think I benchmarked andy's one with CPython
23:53:54 <andythenorth> every nail looks like .join() to me :P
23:53:58 <LordAro> there's probably some python thing that will benchmark it for you :p
23:54:32 <FLHerne> LordAro: Frankly, I don't care, because it's ugly as hell :P
23:54:44 <andythenorth> I'm probably the only author doing nml -> nfo -> grfcodec
23:54:57 <FLHerne> If current speed with CPython is `n`
23:55:13 <FLHerne> Current speed with PyPy is `5n`
23:55:45 <LordAro> yes, but where n is very small for almost all usecases
23:55:46 <FLHerne> Speed with my patch and CPython is about `1.1n`, and with PyPy about `0.5n`
23:56:19 <FLHerne> With andy's, it's `???` and `~n` respectively
23:56:43 <FLHerne> For sane grfs, the perf is irrelevant
23:56:58 <LordAro> like i say, optimising for an edgecase compiler and an edgecase usage is not high on my priority list
23:57:10 <LordAro> i'd rather focus on other areas :p
23:57:13 <FLHerne> For FIRS, on my extremely slow laptop, it goes from 105 seconds to 10 seconds
23:57:22 <FLHerne> Which is non-irrelevant
23:57:58 <FLHerne> (tbf, the rest of FIRS compile still takes a minute or so)
continue to next day ⏵