Log for #openttd on 15th April 2019:
Times are UTC Toggle Colours
00:01:05  <pnda> Eddi|zuHause, well, most are written in C, which I have no idea of, and I hate all of their GUIs. There's one or two web apps like thelounge which are quite good, but I'd like to make my own.
00:02:42  <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick commented on issue #7513: OpenTTD closes itself with no crash dump
00:03:58  <pnda> Just a quick question, what's a good loading_speed for a train? Is 38 a lot?
00:05:43  <Eddi|zuHause> i would do something like capacity/4 or so
00:06:55  <Samu> nerf trains
00:08:53  <pnda> huh maybe I have always had my trains load slowly.... Had a few of them at around 10
00:09:06  <pnda> Maybe the complaints on tt-forums were real
00:14:46  <Samu> how to interpret that call stack?
00:16:20  <Samu> - it got fed up of doing the same thing, it decided to crash?
00:32:12  <glx> looks like a loop in garbage collection
00:32:48  <glx> squirrel probably allocated more memory than available at some point
00:36:25  <glx> AI creates too many and too big arrays I think (but expected with this map size and the number of towns and industries I guess)
00:39:39  *** cHawk has joined #openttd
00:39:42  <DorpsGek_II> [OpenTTD/OpenTTD] glx22 commented on issue #7513: OpenTTD closes itself with no crash dump
00:41:32  <Samu> how much memory is available, do you know? I always notice it's around the 4GB mark
00:41:32  *** rocky11384497 has quit IRC
00:41:37  <Samu> that it happens
00:42:14  *** rocky11384497 has joined #openttd
00:44:50  <glx>
00:44:57  *** pnda has quit IRC
00:45:06  <glx> unlimited unless it's a 32bit exe
00:47:18  <Samu> it was 64bit debug build
00:48:02  <Samu> how much for the virtual machine?
00:48:09  <Samu> squirrel is a virtual machine, i hear
00:48:26  <glx> it's a VM inside openttd
00:51:08  <glx> but visual studio is still 32bit
00:53:21  <Samu> oh :(
00:55:03  <glx> debugger is 64bit
01:00:00  <glx> but in this case I think the stack is the limit
01:27:42  *** Supercheese has quit IRC
01:28:03  *** Supercheese has joined #openttd
01:33:49  <DorpsGek_II> [OpenTTD/OpenTTD] glx22 updated pull request #7509: Codechange: replace left QSortT() with std::sort()
01:53:15  <DorpsGek_II> [OpenTTD/OpenTTD] glx22 updated pull request #7509: Codechange: replace left QSortT() with std::sort()
01:56:41  *** HerzogDeXtEr has quit IRC
02:10:22  *** Flygon has joined #openttd
02:18:09  <DorpsGek_II> [OpenTTD/OpenTTD] stale[bot] closed issue #7188: Both server and client crash, AI related
02:22:12  <Samu> dang, random deviation and random range occurring when in-game, I set up scripts
02:23:06  <Samu> I'm not sure if that's okay
02:24:10  <Samu> they're not started as random, part of me says that's not okay
02:25:04  <Samu> i've just selected them from a list of available scripts, and their config is already loaded with the parameters randomized
02:25:17  *** Smedles has quit IRC
02:26:23  <Samu> i mean, the script didn't start yet, i still can reset parameters to their defaults
02:27:08  <Samu> I suppose it is okay to randomize in this case, it's just that the behaviour is not consistent with its definition.
02:31:35  *** D-HUND has joined #openttd
02:34:53  *** debdog has quit IRC
02:37:23  *** Smedles has joined #openttd
02:41:28  *** Samu has quit IRC
02:41:49  *** Smedles has quit IRC
02:43:01  *** Smedles has joined #openttd
02:49:12  *** Smedles_ has joined #openttd
02:52:32  *** Smedles has quit IRC
02:57:32  *** Smedles_ has quit IRC
03:37:46  *** glx has quit IRC
04:59:00  *** Supercheese has quit IRC
04:59:21  *** Supercheese has joined #openttd
05:46:01  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7270: Introduce CMake (and removing all other project-related code)
06:31:07  *** Smedles has joined #openttd
06:42:42  *** andythenorth has joined #openttd
07:25:56  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #7512: Replace SmallStackSafeStackAlloc and AutoFreePtr
07:58:52  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on issue #7513: OpenTTD closes itself with no crash dump
08:09:46  *** Supercheese has quit IRC
08:37:04  *** andythenorth is now known as Guest284
08:37:04  *** andythenorth has joined #openttd
08:38:55  *** Guest284 has quit IRC
08:40:23  *** andythenorth has quit IRC
09:07:04  <peter1138> Hmm
09:08:04  <Artea> bah
09:08:10  <peter1138> moo?
09:08:15  <Artea> I reinstalled CentOS 7 in VPS
09:08:25  <Artea> is giving me errors compiling UnrealIRCd
09:08:34  <Artea> :(
09:08:45  <Artea> I was so fire up yesterday
09:08:50  <peter1138> Oh
09:08:54  <Artea> thinking I would make my test irc server
09:08:55  <peter1138> Missing dev packages?
09:08:57  <Artea> :(
09:09:17  <Artea> OpenSSL has been a b**** on CentOS 7
09:09:27  <Artea> is using very old version, 1.0.2k
09:10:02  <Artea> and error that gave me, was due to OpenSSL binary not be installed
09:10:04  <peter1138> See the problem is you are using CentOS, not Debian.
09:10:14  <Artea> yes
09:10:26  <peter1138> But you probably want an openssl dev package, not just the binaries.
09:10:27  <Artea> I tried a long time ago
09:10:37  <Artea> and I liked so I sticked with it
09:10:42  <Artea> but was CentOS 6
09:11:00  <peter1138> I looked after a CentOS machine for a while. It was horrible. I Debianed it eventually.
09:11:02  <Artea> I installed openssl-libs
09:11:12  <peter1138> openssl-libs is not dev packages, I would guess.
09:11:26  <peter1138> openssl-devel?
09:11:29  <Artea> it seems like OS don't know OpenSSL is there
09:12:16  <Artea> it's installed too
09:12:30  <Artea> well
09:12:41  <Artea> I will be back later
09:12:50  <peter1138> Come back with Debian ;)
09:12:58  <Artea> or Ubuntu, no ?
09:13:14  <peter1138> Nah, that's just a 3rd rate imposter.
09:13:20  <Artea> lol
09:13:52  <peter1138> Feel free to ignore me about this, I'm just a little bit biased  :)
09:14:06  <Artea> nah
09:14:18  <Artea> I dont ignore people for talking and helping me out :)
09:15:24  <Artea> bbl
09:16:58  *** Artea has quit IRC
09:31:28  *** andythenorth has joined #openttd
09:31:53  *** andythenorth has left #openttd
09:35:17  *** Artea has joined #openttd
09:35:21  <Xaroth> Ubuntu is fine
09:35:33  *** Artea is now known as _Artea
09:36:46  <_Artea>  Reinstalling the VPS (95%)
09:38:13  <_Artea> I really wanna test out IRC Services under MySQL DB
09:38:31  <_Artea> I miss that so much
09:47:36  <_Artea> YES
09:47:40  <_Artea> it worked
09:47:52  <_Artea> made the SSL cert for UnrealIRCd in CentOS 6
10:10:52  <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh commented on issue #7513: OpenTTD closes itself with no crash dump
10:19:20  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on issue #7513: OpenTTD closes itself with no crash dump
10:22:49  <_Artea> peter1138
10:22:58  <_Artea> I installed UnrealIRCd with SSL support :D
10:23:53  <_Artea> now install the rest of the stuff
11:07:19  *** juzza1 has quit IRC
11:10:54  *** juzza1 has joined #openttd
11:13:51  *** gelignite has joined #openttd
11:16:26  *** Corns343434 has joined #openttd
11:22:20  <Corns343434> Patch idea, map Generation creates roads that connect every town together
11:23:40  <Corns343434> Like in transport fever or Simon
11:23:42  <peter1138> "There's a gamescript for that"
11:23:58  <peter1138> But yeah, if you want to make a patch and PR, go for it.
11:24:04  <Corns343434> Simutrans*
11:24:30  <Corns343434> Wait is there actually a game script for it already? I know about the AI that does it kinda
11:24:37  <DorpsGek_II> [OpenTTD/OpenTTD] JGRennison commented on issue #7513: OpenTTD closes itself with no crash dump
11:45:39  <peter1138> Not sure, I thought there was.
11:57:06  *** Corns has joined #openttd
11:59:31  *** Corns has left #openttd
12:02:12  *** Corns343434 has quit IRC
12:06:56  *** Artea has joined #openttd
12:19:18  *** _Artea has quit IRC
12:27:44  <Artea> I'm tired
12:28:53  <Artea> but already got stuff sorted out :)
13:08:45  *** Samu has joined #openttd
13:15:41  <Samu> hi
13:21:03  <Artea> hi
13:27:01  <Samu> I've been thinking about this line
13:28:05  <Samu> check again in a year can be at times annoying
13:34:08  <Samu> making it checking daily can also be annoying, if the start_date of the next ai is also a big number
13:35:49  <Samu> how to solve... hmmm thinnk
13:36:01  <Xaroth> monthly check?
13:37:45  <Samu> it should check the start_date of the AI once an AI can start and reset the counter
13:42:42  *** nielsm has joined #openttd
13:43:20  <Samu> it should return a special number that is to be treated as "unable to check at the moment"
13:44:06  <Samu> once it gets a value that is not the special number, it then sets the daily countdown
13:55:56  <Samu> _next_competitor_start is a uint
13:56:23  <Samu> GetStartNextTime is a int
13:57:33  <Samu> what will happen if GetStartNextTime returns UINT_MAX?
13:58:10  <Samu> what exactly is returned?
14:00:38  <Samu> -1
14:01:09  *** Flygon has quit IRC
14:04:49  <Samu> -1 is then turned to 1 tick
14:04:51  <Samu> meh
14:05:07  <Samu> can't do that
14:05:10  <Samu> i dont like it
14:08:15  <Samu> @calc 0xffffffff / 74
14:08:15  <DorpsGek> Samu: 58040098.5811
14:19:07  *** Corns has joined #openttd
14:20:40  <Corns> hello update: there is no city-road building gamescript
14:22:48  <Corns> author is on hiatus
14:22:48  <Corns>
14:25:03  <Samu>
14:25:55  <Corns> oh hey thanks :)
14:28:25  <Samu> oh snap, CityConnected is an AI
14:28:34  <Samu> i gave u a GS lib
14:30:03  <Samu> i fail at helping
14:31:48  <Samu> it only lists version 3 :(
14:32:14  <Samu> no idea how to download v2 then
14:32:15  <Corns> oh LOL
14:32:27  <Corns> dw i'm just reverse engineering it for a PR
14:34:20  <Artea> grrrr
14:34:27  <Artea> can't setup sendmail :(
14:34:33  <Artea> I sux configuring VPS
14:37:34  <Samu> oh, INT_MAX is also a thing!
14:37:36  <Samu> cool
14:38:01  <Samu> solves needlessly conversions from int to uint
14:38:11  <DorpsGek_II> [OpenTTD/OpenTTD] glx22 commented on pull request #7270: Introduce CMake (and removing all other project-related code)
14:38:25  <Corns> were you relying on uint = -1
14:40:16  <Samu> 	if (_next_competitor_start == 0 || _next_competitor_start == (INT_MAX / DAY_TICKS) * DAY_TICKS - 1) {
14:41:15  <Samu> - 1 would make it check daily
14:41:37  <Samu> if (_next_competitor_start == 0 || _next_competitor_start == (INT_MAX / DAY_TICKS) * DAY_TICKS - DAY_TICKS) {
14:41:41  <Samu> maybe this is better
14:41:57  <Samu> oops
14:42:02  <Samu> -1 wouldn't*
14:42:28  *** glx has joined #openttd
14:42:28  *** ChanServ sets mode: +v glx
14:42:42  *** Wormnest has joined #openttd
14:44:18  <Samu> @calc INT_MAX / 74
14:44:18  <DorpsGek> Samu: Error: There's really no reason why you should have underscores or brackets in your mathematical expression.  Please remove them.
14:44:33  <Samu> @calc 2147483647 / 74
14:44:33  <DorpsGek> Samu: 29020049.2838
14:44:57  <Samu> @calc 29020049 * 74
14:44:57  <DorpsGek> Samu: 2147483626
14:45:15  <Samu> @calc 21474863626 - 74
14:45:15  <DorpsGek> Samu: 21474863552
14:47:13  <Samu> @calc 2147483626 - 74
14:47:14  <DorpsGek> Samu: 2147483552
14:50:39  <milek7> Artea: why sendmail?
14:50:55  <peter1138> Probably doesn't know any better.
14:51:05  <peter1138> Artea, don't use sendmail, simply.
14:53:19  <Artea> yeah
14:53:25  <Artea> I need a break from VPS now
14:53:26  <Corns> Are commit messages supposed to be typed in present tense?
14:53:36  <Artea> I think I going screw the things up if I continue
14:53:53  <Artea> at least have my IRCd working with SSL
14:54:44  <milek7> use exim or postfix instead of sendmail
14:56:50  <peter1138> postfix++
14:58:45  *** Thedarkb-T60 has joined #openttd
15:00:11  <Artea> yeah
15:00:11  *** Wormnest has quit IRC
15:00:19  <Artea> I was thinking of Postfix
15:00:27  <Artea> but dont know how to config
15:00:35  <Artea> and this CentOS 7
15:00:37  <Artea> GRRRRRRR
15:00:51  <Artea> I like it
15:00:59  <Artea> but is a pain in the backhead
15:01:37  *** Wormnest has joined #openttd
15:23:20  *** WWacko1976-work has quit IRC
15:34:04  *** acklen has quit IRC
15:34:31  <Samu> i need an english expert
15:34:43  <Samu> 	/* Currently no AI can be started. Return an indefinited number of days. */
15:34:44  <Samu> 	return INT_MAX / DAY_TICKS;
15:34:51  <Samu> indefinited?
15:35:00  <nielsm> that's not a word
15:35:08  <Samu> unspecified?
15:35:55  <nielsm> infinite, indefinite, undefined
15:35:59  <Samu> unknown? indeterminate? unsettled? unfixed?
15:36:11  <Samu> undefined
15:36:12  <Samu> ok
15:36:30  <nielsm> but undefined is a tricky word
15:36:36  <Samu> undetermined
15:36:48  <nielsm> because in programming context it tends to have a specific meaning
15:37:08  <nielsm> "anything can happen/any value is possible, there is no rule to it"
15:37:30  <nielsm> also is this a sentinel value that will need to be compared against?
15:38:17  <Samu> yes
15:38:54  <Samu> if this value is set, code handles it differently than a real value
15:39:02  <nielsm> if this is a value you'll actually need to check "is it the sentinel value" then make it a defined constant
15:39:54  <nielsm> and if it may need to be transmitted over the network in some way, make sure it does not depend on compiler-defined values such as INT_MAX
15:40:19  <Samu> hmm, can't be INT_MAX then?
15:40:37  <Samu> well, some value higher than 3600
15:40:54  <Samu> or 3600 * DAY_TICKS
15:40:59  <nielsm> it's dangerous to assume that INT_MAX has the same value on any compiler
15:41:01  <Samu> have to check
15:41:03  *** HerzogDeXtEr has joined #openttd
15:42:25  <Samu> max permitted value for start_date is 3600
15:42:35  <Samu> so, anything above that range
15:47:43  <Samu> return AI::START_NEXT_MAX + DAY_TICKS;
15:47:47  <Samu> 3600+74
15:47:52  <Samu> let me see
15:58:41  <Samu> crap, this is already days
15:58:44  <Samu> return AI::START_NEXT_MAX + 1;
15:59:22  <Samu> i was adding 74 days lol
16:01:24  *** Corns has quit IRC
16:04:25  <Samu> does it still warrant a defined constant? I think not
16:04:40  <Samu> if (_next_competitor_start == 0 || _next_competitor_start == AI::START_NEXT_MAX * DAY_TICKS) {
16:04:53  <Samu> return AI::START_NEXT_MAX + 1;
16:04:58  <Samu> that's how it's being used
16:05:25  <Samu> _next_competitor_start = max(1, AI::GetStartNextTime() * DAY_TICKS);
16:06:12  <Samu> if (_game_mode != GM_MENU && AI::CanStartNew() && --_next_competitor_start == 0) {
16:06:35  <Samu> it will never reach 0
16:06:51  *** acklen has joined #openttd
16:07:32  <Samu> it won't use the YEARLY check counter
16:07:46  <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh opened pull request #7514: Scale oil refinery edge distance up by map size
16:08:46  <Samu> every 74 ticks it rechecks if there's an AI that can start
16:09:11  <Samu> if there is, it first restarts the counter with that AI's start_date value
16:09:52  <Samu> and now the value is no longer going to be 3600, so, it will reach 0, which is when the company starts
16:10:48  *** Thedarkb1-X40 has joined #openttd
16:11:11  *** Thedarkb1-T60 has joined #openttd
16:13:45  <Samu> what it was doing was: check every 365 * 74 ticks if it could start next AI. And once that counter reached 0, it would try to start an AI, without caring about the start_date
16:14:34  <Samu> if it couldn't start, another 365*74 countdown was started
16:14:48  *** Thedarkb-X40 has quit IRC
16:14:53  *** Thedarkb-T60 has quit IRC
16:18:54  <Artea> that is why AI dont start if start_date = 1 ?
16:21:23  <Samu> it happens when there's no more company slots for AIs to start
16:21:30  <Samu> so, 15 companies
16:22:03  <Samu> if then a company is removed, the counter was not reset
16:22:12  <Samu> it still isn't, but now it's checked daily
16:22:42  <Samu> instead of ... having to wait for the remaining countdown
16:22:53  <Samu> which could be anywhere from 1 to 365 * 74
16:24:09  <Artea> I not sure
16:24:47  <Artea> AIAI = start_date=1
16:24:54  <Artea> isnt by days ?
16:25:13  <Samu> well, there is random deviation
16:25:25  <Samu> can add up to +60 to that
16:25:36  <Samu> it's a bit complex
16:25:58  <Samu> but the problem i'm solving is not that
16:26:31  <Samu> it's only really happening when there's no more room for extra companies
16:30:00  *** sla_ro|master has joined #openttd
16:32:40  <Artea> odd
16:33:54  *** Supercheese has joined #openttd
16:34:49  *** Progman has joined #openttd
16:43:23  <Samu> i'm so bad at comments, i'm gonna need help
16:48:49  <Samu> crap, i submited to master, stupid me
16:48:59  <Samu> how to undo?
16:51:06  <Samu> rebase -i drop ?
16:54:06  <Samu> omg i can't believe i commited the same mistake twice
16:54:08  <Samu> :(
16:54:57  <Artea> :S
16:55:43  <Artea> How to "undo" rebasing commits ?
16:55:53  <nielsm> use the reflog
16:55:55  <glx> only one commit ?
16:56:09  <glx> git reset HEAD^
16:56:20  <glx> keep the changes
16:56:23  <Samu> i went with rebase -i and dropped :(
16:56:27  <nielsm> "git reflog" shows all the commit hashes that have been HEAD so you can reset your branches to one of those
16:56:43  <rubenwardy> git reset --hard HEAD~1
16:56:56  <glx> no --hard removes the changes
16:57:14  <rubenwardy> or soft
16:57:20  <nielsm> uh git reset HEAD^ will just remove the top commit, it won't undo whatever a rebase did
16:57:27  <nielsm> it won't get you dropped commits back
16:57:51  <nielsm> nor will it un-squash commits
16:58:29  <glx> but it's enough if there's only one commit
16:58:51  <nielsm> no because it will just remove the top commit
16:58:55  <nielsm> it won't undo the rebase
16:59:05  <nielsm> it won't restore the old tree
16:59:15  <Samu> i typed git reflog, how to get out of this?
16:59:20  <nielsm> q
16:59:20  <Samu> ctrl-c ?
16:59:22  <Samu> ah
17:00:42  <Eddi|zuHause> the reflog will tell the ID where the rebase started, then use reset with that ID
17:03:28  <nielsm> wtf
17:03:30  <nielsm> 1>------ Build started: Project: version, Configuration: Debug Win32 ------
17:03:30  <nielsm> 1>Error HRESULT E_FAIL has been returned from a call to a COM component.
17:03:35  <nielsm> and then my entire build fails
17:04:02  <Samu> ah, i had that when i updated visual studio
17:04:19  <Samu> i solved it by removing ... uhm... some files
17:04:26  <Samu> forgot where
17:05:39  <Samu> ah,
17:05:40  <Samu>
17:07:02  <Samu> v16 for me
17:07:05  <Samu> not v15
17:07:47  <nielsm> hmm yeah removing the suo did fix it
17:15:40  *** Wolf01 has joined #openttd
17:16:26  <Wolf01> o/
17:25:01  <V453000> arrived to the end of 55k lines of nuts.nml :D ofc I missed some things ._.
17:25:19  <Eddi|zuHause> why would you go through that manually?
17:29:45  <glx> yeah I got this error too when I upgraded VS2019
17:29:49  <nielsm> he's... nuts?
17:30:07  <glx> Sacro: v15 is 2017
17:31:02  <LordAro> Samu: &
17:31:05  <LordAro> ^
17:32:12  <Samu> hmm?
17:33:15  <DorpsGek_II> [OpenTTD/OpenTTD] SamuXarick opened pull request #7515: Fix: Really check for start_date once an AI company slot becomes available.
17:33:33  <Samu> i fail at explanations
17:33:50  <Samu> descriptions, comments, etc...
17:34:00  <Samu> who can help?
17:36:13  <Samu> basically, if there's no slots, it checks daily, using the interval 3601 - 3600
17:36:24  <Samu> and never reaches 0
17:37:24  <Samu> if it gets a start_date lower than that 3601-3600 interval, then the real countdown starts, using that start_date value. When it reaches 0, it tries to make ai company
17:39:17  <glx> say that in the PR :)
17:40:31  <peter1138> Hmm, cold fnigers
17:41:35  <Samu> i could make it check every tick
17:42:05  <Samu> for now it's checking every 74 ticks
17:42:19  <Samu> which is what truebrain likes
17:43:34  <Samu> _next_competitor_start == (AI::START_NEXT_MAX + 1) * DAY_TICKS - 1 to check every tick
17:45:10  *** HerzogDeXtEr has quit IRC
17:58:22  * peter1138 fudges with road removal/conversion costs.
18:00:59  <peter1138> Heh, AIs running trams all over the place instead of trains.
18:03:04  * peter1138 blasts a bit of Foxtrot for old-times sake.
18:03:09  <peter1138> @seen belugas
18:03:09  <DorpsGek> peter1138: belugas was last seen in #openttd 2 years, 2 weeks, 0 days, 21 hours, 44 minutes, and 37 seconds ago: <Belugas> yeah, a birthday :)  a good friend of mine!
18:03:15  <peter1138> :/
18:04:20  * nielsm works on another try at keeping track of squirrel allocations
18:09:13  * peter1138 ponders dinner
18:10:45  <LordAro> peter1138: where's the bike?
18:12:41  <peter1138> At the shop. Gear cable snapped and I cba doing it myself.
18:13:16  <peter1138> These new-fangled brifters and internal cable routing are too much for me :p
18:13:33  <peter1138> I can cope with down-tube friction shifters.
18:15:01  <Samu> cat !
18:15:07  <peter1138> WAITING FOR BATTLE
18:15:17  <peter1138> [[epic instrumental-section]]
18:15:26  *** APTX has quit IRC
18:16:31  *** APTX has joined #openttd
18:17:08  <LordAro> peter1138: oh dear
18:17:10  <peter1138> I wonder, should road and tram vehicles be separated in the UI?
18:18:13  <peter1138> Nothing major, I decided to walk instead of drive so I can pig out on more food ;)
18:18:34  <LordAro> :)
18:19:26  <nielsm>
18:19:34  <nielsm> something happens!
18:19:45  <peter1138> Memory!
18:20:03  <peter1138> Nice work.
18:20:09  *** andythenorth has joined #openttd
18:20:17  <peter1138> Uh oh, this guy again
18:20:30  <nielsm> I'm not sure if it context switches the allocator correctly
18:20:53  <nielsm> but at least it measures something :P
18:22:06  <peter1138> I got as far as simple counting in the console, but had some weird values.
18:24:59  <nielsm> now also featuring GS
18:25:27  <peter1138> Now run one of Samu's AIs :p
18:25:47  <Samu> ohhh, memory!
18:26:40  <peter1138> I suppose you *could* count the memory used by vehicles etc... but... eh... no.
18:27:16  <peter1138> That's simple anyway, count * sizeof train etc
18:27:30  <nielsm> well yeah
18:27:57  <nielsm> unless we let newgrf allocate arbitrary memory :P
18:28:25  <peter1138> Those varaction chains add up ;)
18:30:42  <V453000> Eddi|zuHause: adding support for 1 cargo to all vehicles :D and let's say the code isn't exactly clean.
18:31:20  <nielsm>
18:31:22  <nielsm> there's a branch
18:34:44  <nielsm> make a PR of it yet?
18:35:49  <Artea> damn
18:35:59  <Artea> for who adviced not use sendmail
18:36:08  <Artea> I'm busted
18:36:18  <Artea> seems I need to use it in IRC Services
18:40:45  <DorpsGek_II> [OpenTTD/OpenTTD] nielsmh opened pull request #7516: Custom allocators for Squirrel
18:44:17  <Samu> cool
18:44:24  <Samu> gonna test
18:45:54  <peter1138> Artea, no, you don't need *sendmail*
18:46:29  <Artea> aiiii
18:46:32  <Artea> peter
18:46:52  <Artea> I going need for Anope IRC Services apparently
18:47:00  <milek7> 'sendmail - Postfix to Sendmail compatibility interface'
18:47:10  <Artea> or really learn how to send mails via services
18:47:14  <milek7> The Postfix sendmail(1) command implements the Postfix to Sendmail compatibility interface.
18:48:41  <Artea> milek: I'm building a test irc server
18:48:45  *** circ-user-nH1UG has joined #openttd
18:49:02  <DorpsGek_II> [OpenTTD/OpenTTD] stormcone opened pull request #7517: Fix: Fix typos.
18:49:22  <Artea> on services the mail uses sendmail -t
18:49:24  <nielsm> what you need is not the specific software package called "sendmail", what you need is something implementing the correct API that is compatible with sendmail
18:49:48  <Artea> I have no clue what need to do
18:49:54  *** circ-user-nH1UG is now known as supermop_work
18:49:57  <Artea> brb
18:51:25  <peter1138> Ok, that's dinner done.
18:51:29  <peter1138> Now for dessert...
18:51:51  <nielsm> I gave one of my birds a bit of pizza and I think he liked it (but he also wanted to get back to sleepytime)
18:51:51  <Artea> sorry
18:52:02  <Artea> having issues connecting my services to server
18:54:09  <Samu> ewww really slow
18:54:19  <Samu> were u testing with 10k ops?
18:54:25  <Samu> 250k ops too slow now
18:55:33  <Samu> hehe, NoCab eats memory
18:57:49  <nielsm> I did not test with any particular settings, I did not test for performance, I only did smoke testing for "looks not entirely incorrect"
18:59:35  *** crem1 has joined #openttd
19:00:33  * andythenorth ponders desert
19:00:38  <andythenorth> peter1138: what was dinner?
19:00:51  <Samu> waiting for NoCAB to crash/poof openttd
19:00:56  <Samu> unless you fixed that
19:01:08  <DorpsGek_II> [OpenTTD/OpenTTD] michicc commented on pull request #7512: Replace SmallStackSafeStackAlloc and AutoFreePtr
19:01:16  *** crem has quit IRC
19:01:16  <peter1138> andythenorth,
19:01:38  <andythenorth> that's not bad
19:01:45  <andythenorth> square plates improve all food
19:01:59  <peter1138> Yes.
19:02:01  * andythenorth had kebabs
19:02:37  <peter1138> Although it's actually a square cereal bowl. Portion control, and keeps things from falling off :p
19:03:00  <peter1138> Eh, probably a soup bowl, that'd be a lot of cereal.
19:03:38  <peter1138> So road conversion costs, eh?
19:03:54  <peter1138> Can I just make it cost the same as removal & build?
19:04:12  <peter1138> It's not like rails where you can reuse or sell the rails...
19:04:16  <peter1138> Although for trams? Hmm...
19:04:19  <andythenorth> removal and build
19:04:25  <andythenorth> oof trams
19:04:56  <peter1138> The money bug currently is that you get the removal cost... *back*... whoever came up with that idea, I dunno.
19:06:24  <Samu> I agree, nerf trains
19:06:55  <Samu> eh, openttd poofed, when I wasn't here at the desk
19:11:44  <andythenorth> nerf everything
19:14:03  <nielsm> I should derail the valley a bit again some time soon, they apparently fixed my number one complaint now: trees and rocks on the tracks
19:14:09  *** raphikov has joined #openttd
19:14:49  *** Gustavo6046 has quit IRC
19:19:51  <Samu> boom, poofed at 3.38 GB memory
19:26:44  <peter1138> What are you running 32 bit?
19:29:59  <Samu> no
19:30:04  <Samu> 64-bit
19:30:42  <Samu> release x64
19:30:51  <peter1138> nielsm, hmm, any way to do the allocator without the indirection through function pointers?
19:31:23  <nielsm> I doubt it
19:31:28  <peter1138> Really?
19:32:27  <nielsm> well could just remove sqmem.cpp and implement those functions somewhere outside squirrel
19:32:46  <glx> 32 or 64bit, the issue is in the garbage collector I think
19:32:47  <nielsm> and not extend the squirrel api with sq_setallocator at all
19:33:30  <glx> but AI can allocate way too more memory ;)
19:34:24  <glx> *much
19:34:27  <peter1138> nielsm, yeah, nielsm's PR could allow a memory limit per VM.
19:34:36  <nielsm> lol
19:35:31  <nielsm> also need to check how squirrel deals with an allocation failing...
19:35:37  <glx> even in your video, a starting AIAI use twice the memory of the AI started before it
19:35:53  <nielsm> maybe allocation failure needs to be handled by an exception bubbling up through the entire VM?
19:36:30  <milek7> is allocation failure even a thing?
19:36:39  <milek7> or it just oom-kills
19:36:51  <nielsm> if you put each squirrel VM in its own arena it can be a thing
19:37:11  <Samu> i'm imagining 240 nocabs on a 4096x4096 map
19:37:14  <nielsm> where one AI runs out of memory allowed and fails allocating, while the rest of the game is fine
19:37:22  <peter1138> Yeah, return nullptr; might cause breakage :)
19:38:31  <peter1138> Hmm, maybe I should spreadsheet this build/sell/convert cost bullshit.
19:38:52  <peter1138> Oh, I can just printf on init ;)
19:41:40  <Samu> @calc 3.38 * 240
19:41:40  <DorpsGek> Samu: 811.2
19:41:54  <Samu> I need 811 GB of ram
19:43:40  <Samu> maybe one day in the future, 1 TB of ram will be the norm
19:43:52  <glx> Samu: tried to crash with 2 AIs ?
19:44:11  <Samu> no, only with 1
19:44:58  <glx> try with 2, to see if the fails happen on same memory level
19:45:25  <Samu> yesterday I tried 15 NoCABs, turns out I dont have enough ram for it
19:45:29  <Samu> didn't crash
19:45:38  <Samu> but the system was already too slow to cope
19:45:52  <peter1138> glx, I think he's in write-only mode again.
19:46:08  <Samu> Heh, ok let me try 2
19:46:17  <peter1138> Yes, we'll let you do that o_O
19:49:28  <Samu> started
19:50:46  <Samu> i was wondering, just disable the garbage collector
19:50:51  <Samu> fix all problemos!
19:51:26  <glx> no because without it the memory is not reused
19:52:00  *** Thedarkb1-T60 has quit IRC
19:55:16  <Samu>
19:55:22  <Samu> almost there
19:55:32  <peter1138> Hmm, well, nothing is negative now.
19:56:11  <peter1138> "SamuPatchPack" ... that's a terrifying idea.
19:56:30  <Samu> it was your idea
19:56:54  <peter1138> I doubt it.
19:57:38  <Samu> strange
19:58:00  <Samu> memory keeps rising in task manager, but over there in openttd it's still 3.18 GB for each
19:58:11  <nielsm> oh well, I delivered some stuff in derail valley without derailing, enjoyed running on clear rails, and then teleported over to old bob's garage and unlocked it
19:58:25  <nielsm> and yes his secret vehicle is not controllable with mouse and keyboard
20:01:15  <peter1138> Heh
20:01:49  <andythenorth> hmm
20:03:40  *** Thedarkb-T60 has joined #openttd
20:04:10  <Samu> interesting, no crash yet
20:04:11  <andythenorth> what was I doing?
20:04:44  <Samu> they reached the point they built vehicles
20:05:00  <Samu> aha, crash
20:05:06  <Samu> spoke too early
20:05:23  <peter1138> andythenorth, deciding how much road should cost to remove.
20:05:42  <peter1138> Should highway cost the same as dirt tracks to remove?
20:05:46  <glx> so same memory level ? more ? less ?
20:06:11  <DorpsGek_II> [OpenTTD/OpenTTD] michicc approved pull request #7517: Fix: Fix typos.
20:06:44  <DorpsGek_II> [OpenTTD/OpenTTD] michicc merged pull request #7517: Fix: Fix typos.
20:07:14  <andythenorth> peter1138: I usually just level all that crap
20:07:20  <andythenorth> they're not interesting costs
20:07:36  <peter1138> andythenorth, not interesting, but it affects road conversion cost.
20:07:39  <Samu> aha here it is
20:07:40  <Samu>
20:07:45  <peter1138> andythenorth, which is currently exploitable.
20:07:57  <andythenorth> is it exploitable if they all cost same to remove?
20:08:24  <glx> Samu: but with double memory this time it seems
20:08:24  <peter1138> andythenorth, not really. The exploit is that currently conversion can give you more money than you spent in the first place.
20:08:33  <peter1138> andythenorth, which is obviously wrong.
20:08:55  <andythenorth> if it was left to me, cost to clear would be same in all cases
20:09:06  <peter1138> Well I'm asking you.
20:09:07  <andythenorth> and would just use the default cost to clear road
20:09:12  <andythenorth> as prior to NRT
20:09:19  <peter1138> Yes, and the conversion cost?
20:09:37  <andythenorth> sum of cost of clearing + cost of building
20:09:47  <peter1138> So zero benefit. Okay.
20:09:55  <andythenorth> zero benefit
20:10:32  <andythenorth> tram tracks are slightly awkward, because realism + precedent with trains
20:11:19  <peter1138> It's possible to have different behaviour between roads and trams.
20:11:44  <Samu> gonna try borkai, think it also hoards memory like crazy
20:14:16  <Samu> afk dinner
20:17:00  *** sla_ro|master has quit IRC
20:20:31  <nielsm> gn, I'll do some more with that memory management tomorrow, probably
20:21:45  <peter1138> Byte
20:21:46  <peter1138> -t
20:21:59  <andythenorth> I should probably be doing some newgrf or something
20:22:07  <peter1138> Or testing NRT.
20:22:13  <peter1138> Finding the bugs.
20:23:45  <peter1138> Hmm.
20:24:05  <peter1138> Flat fee for roads, variable for tram tracks.
20:24:53  <peter1138> But never beneficial to build one type and then convert.
20:26:36  <peter1138> Hmm, of course, AIs only ever build original ROAD and ELRL.
20:28:36  *** nielsm has quit IRC
20:32:49  *** Progman has quit IRC
20:38:06  <Artea> yeahhhhh
20:38:15  <Artea> compiling OTTD in VPS :D
20:38:34  <andythenorth> I should make new Road Hog
20:38:38  <andythenorth> to test NRT
20:38:42  <andythenorth> or HEQS or something
20:39:28  <andythenorth> or I should finish Iron Horse, Unsinkable Sam and FIRS
20:39:34  <andythenorth> before wrecking anything new :P
20:40:23  *** arikover has joined #openttd
20:42:52  <andythenorth> any Eddi|zuHause? o_O
20:44:57  <peter1138> I should find my soldering iron.
20:45:15  <andythenorth> you been watching Big Clive again?
20:45:38  <peter1138> Nah.
20:45:55  <peter1138> Building my freezer temperature sensor :p
20:46:33  <peter1138> Hmm, removing tram tracks now returns some money.
20:47:05  <peter1138> Which is kinda odd :/
20:47:39  <peter1138> But at least you never make money now.
20:48:39  <dwfreed> scrap?
20:49:03  <peter1138> Something like that.
20:49:05  <dwfreed> RCT had the same behavior; you'd get some money back from demolishing a ride
20:49:23  <peter1138> Normal rail track already does it.
20:52:33  <peter1138> Oh, silly road types. This one changes visual style based on date.
20:52:34  <peter1138> D'oh.
20:54:37  <Samu> back
20:54:42  <Samu> borkai didn't crash yet
20:54:52  <Samu> it's sitting at 800 MiB
20:56:41  <glx> <@peter1138> Oh, silly road types. This one changes visual style based on date. <-- desync possible no ?
20:57:02  <peter1138> No?
20:57:14  <peter1138> varaction date
20:57:20  <peter1138> Not load date
20:57:24  <glx> ah it's at runtime
21:03:23  *** andythenorth is now known as Guest320
21:03:56  *** andythenorth has joined #openttd
21:03:59  <andythenorth> ouch
21:04:05  <andythenorth> laptop hard crashed :(
21:04:14  <andythenorth> strictly speaking, I did kill -9 the windowserver
21:04:23  <andythenorth> but why should that crash it? :P
21:06:04  *** gelignite has quit IRC
21:06:46  *** Guest320 has quit IRC
21:10:18  <peter1138> Well...
21:17:06  <peter1138> Okay, regression test still passes.
21:17:10  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #6811: Feature: Add NotRoadTypes (NRT)
21:19:11  *** Wolf01 has quit IRC
21:20:01  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN commented on pull request #6811: Feature: Add NotRoadTypes (NRT)
21:20:36  <peter1138> Oh, I need to rebase due to conflicts... in a comment :p
21:22:14  <andythenorth> oof
21:22:23  <andythenorth> also bed :P
21:22:24  *** andythenorth has quit IRC
21:22:25  <peter1138> Nah, easy fix.
21:22:28  <peter1138> BYE THEN
21:23:45  <DorpsGek_II> [OpenTTD/OpenTTD] PeterN updated pull request #6811: Feature: Add NotRoadTypes (NRT)
21:51:58  <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7512: Replace SmallStackSafeStackAlloc and AutoFreePtr
21:52:35  <DorpsGek_II> [OpenTTD/OpenTTD] LordAro approved pull request #7512: Replace SmallStackSafeStackAlloc and AutoFreePtr
21:52:53  <DorpsGek_II> [OpenTTD/OpenTTD] LordAro merged pull request #7512: Replace SmallStackSafeStackAlloc and AutoFreePtr
21:57:31  *** Thedarkb1-T60 has joined #openttd
21:57:43  <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7515: Fix: Really check for start_date once an AI company slot becomes available.
21:58:19  <DorpsGek_II> [OpenTTD/OpenTTD] LordAro commented on pull request #7510: Emscripten support
22:00:00  *** Thedarkb-T60 has quit IRC
22:10:13  *** Gustavo6046 has joined #openttd
22:18:38  *** arikover has quit IRC
22:27:43  *** HerzogDeXtEr has joined #openttd
22:29:08  <Samu> start_date is a hack in itself, maybe
22:29:11  <Samu> kek
22:38:42  <Samu> start_date could become a global parameter, similar to difficulty preset currently is
22:40:26  <Samu> what do you think?
23:02:51  *** tokai has quit IRC
23:06:10  *** tokai has joined #openttd
23:06:10  *** ChanServ sets mode: +v tokai
23:15:11  <Samu> nielsm, got a crash
23:15:20  <Samu> FramerateWindow::DrawElementAllocationsColumn + 230 (D:\OpenTTD\OpenTTD GitHub\OpenTTD\src\framerate_gui.cpp:606)
23:19:39  <Samu> it was trying to GetAllocatedMemory() of a dead instance
23:19:54  <Samu> ai that crashed
23:29:43  <glx> it's just a quick test, needs more work
23:46:03  *** Thedarkb1-T60 has quit IRC
23:52:46  *** Thedarkb1-T60 has joined #openttd
23:56:27  *** Supercheese has quit IRC

Powered by YARRSTE version: svn-trunk