Log for #openttd on 11th May 2020:
Times are UTC Toggle Colours
00:14:20  <DorpsGek_III> [OpenTTD/OpenTTD] BastianInuk commented on issue #7826: Mac OS binaries are unsigned
00:16:54  *** gnu_jj has quit IRC
00:17:19  *** gnu_jj has joined #openttd
00:21:56  *** gelignite has quit IRC
00:34:00  <_dp_> "Do something of which I have no idea what it is :P"
00:34:03  <_dp_> nice comment :p
01:06:58  *** mcbanhas_ has quit IRC
01:25:03  *** Foveafoxy has joined #openttd
01:25:40  *** Foveafoxy is now known as Guest24703
02:10:47  *** D-HUND has joined #openttd
02:14:11  *** debdog has quit IRC
02:41:43  *** Flygon has joined #openttd
03:12:50  *** glx has quit IRC
04:01:55  *** Guest24703 has left #openttd
05:31:01  *** Foveafoxy has joined #openttd
05:31:40  *** Foveafoxy is now known as Guest24721
06:02:46  *** sla_ro|master has joined #openttd
06:11:30  *** Guest24721 has quit IRC
06:18:58  *** snail_UES_ has quit IRC
06:22:42  *** Foveafox_ has joined #openttd
06:40:25  *** andythenorth has joined #openttd
07:34:40  *** Foveafox_ has quit IRC
08:02:58  *** iSoSyS has joined #openttd
08:18:00  <DorpsGek_III> [OpenTTD/nml] matthijskooijman commented on pull request #136: Exclude version update code (fixes #112) and dotfiles from release tarballs.
08:18:06  <DorpsGek_III> [OpenTTD/OpenTTD] giordy commented on issue #8038: Fullscreen mode in MacOSX is almost unusable
08:38:30  *** Smedles has quit IRC
08:43:38  <DorpsGek_III> [OpenTTD/OpenTTD] nielsmh commented on issue #7826: Mac OS binaries are unsigned
08:49:05  <DorpsGek_III> [OpenTTD/OpenTTD] andythenorth commented on issue #8038: Fullscreen mode in MacOSX is almost unusable
08:53:18  *** Samu has joined #openttd
08:56:20  *** Foveafoxy has joined #openttd
08:56:49  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on issue #8083: Add currency Unit - Indian  Rupee
08:57:00  *** Foveafoxy is now known as Guest24733
09:03:28  <_dp_> hmmm, I got desync saves and 1 tick before
09:04:04  <_dp_> but diff is kinda weird, there are some different map tiles but they're mostly farmland %)
09:05:02  *** Smedles has joined #openttd
09:08:12  <andythenorth> so
09:08:18  <andythenorth> 64 cargos might not be enough
09:08:18  *** Smedles has quit IRC
09:09:59  *** Smedles has joined #openttd
09:10:57  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on issue #8083: Add currency Unit - Indian  Rupee
09:11:31  <FLHerne> andythenorth: Only because you bundle all the industries into one megagrf, no?
09:11:39  <FLHerne> *economies
09:11:45  <andythenorth> nah, this is just for the chemicals economy
09:11:53  <andythenorth> I have gone overkill on intricate details
09:12:04  <FLHerne> If you didn't do that, it would be less huge, and people wouldn't not know about the non-default ones
09:12:09  <FLHerne> Oh, ok
09:12:16  <FLHerne> This sounds like it'll be fun :D
09:12:20  <andythenorth> it's stupid
09:12:25  <andythenorth> but I'm learning a lot about plastics
09:12:30  <andythenorth> it's really interesting
09:12:40  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on issue #8083: Add currency Unit - Indian  Rupee
09:13:02  <andythenorth> I don't think it will be a great subject for a transport game though
09:13:04  *** cHawk has quit IRC
09:14:20  <andythenorth> it's all tank cars or pellet hoppers
09:14:28  <andythenorth> although there are some crazy tank cars
09:16:53  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 opened pull request #8136: Add: Currency: INR
09:17:53  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on issue #8083: Add currency Unit - Indian  Rupee
09:18:45  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on issue #8083: Add currency Unit - Indian  Rupee
09:19:11  <andythenorth> Horse Tank wagons?
09:20:08  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on issue #8083: Add currency Unit - Indian  Rupee
09:22:37  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on pull request #8136: Add: Currency: INR
09:34:25  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on pull request #8136: Add: Currency: INR
09:40:26  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #8136: Add: Currency: INR
09:40:26  <DorpsGek_III> [OpenTTD/OpenTTD] giordy commented on issue #8038: Fullscreen mode in MacOSX is almost unusable
09:45:27  <FLHerne> andythenorth: Do power-management warnings not override fullscreen on OSX?
09:45:46  <FLHerne> That seems a bit daft
09:54:22  <andythenorth> FLHerne usually they would, but DOSBox uses a particularly aggressive full-screen mode
09:59:22  <_dp_> finally some desync material
09:59:35  <_dp_> this coal mine production is being transported on server but not on clients
10:03:55  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 updated pull request #8136: Add: Currency: INR
10:05:33  *** supermop_Home has quit IRC
10:07:10  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 updated pull request #8136: Add: Currency: INR
10:08:04  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on pull request #8136: Add: Currency: INR
10:15:00  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on pull request #8136: Add: Currency: INR
10:15:07  *** cHawk has joined #openttd
10:18:47  <DorpsGek_III> [OpenTTD/nml] matthijskooijman opened issue #139: Running regression tests litters examples with empty .nmlcache directories
10:23:05  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on pull request #8136: Add: Currency: INR
10:29:06  *** Markk has quit IRC
10:34:44  *** matt21347 has joined #openttd
10:37:11  *** Markk has joined #openttd
10:41:29  <DorpsGek_III> [OpenTTD/nml] FLHerne commented on issue #139: Running regression tests litters examples with empty .nmlcache directories
10:48:07  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on pull request #8136: Add: Currency: INR
10:49:50  <DorpsGek_III> [OpenTTD/nml] FLHerne commented on issue #139: Running regression tests litters examples with empty .nmlcache directories
10:51:27  <DorpsGek_III> [OpenTTD/OpenTTD] FLHerne commented on pull request #8136: Add: Currency: INR
10:57:22  <DorpsGek_III> [OpenTTD/nml] glx22 commented on issue #139: Running regression tests litters examples with empty .nmlcache directories
11:01:44  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on pull request #8136: Add: Currency: INR
11:02:07  <Eddi|zuHause>  <andythenorth> I have gone overkill on intricate details <-- have you considered turning into factorio modding instead? :p
11:04:39  *** Guest24733 has quit IRC
11:12:54  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 updated pull request #8136: Add: Currency: INR
11:31:59  <blathijs> Hm, it seems that nml 0.5.x broke compatibility somewhere. I tried to compile OpenGFX 0.5.5 with nml 0.5.1, and got an MD5 check mismatch in ogfxe_extra.grf. Is that expected, or a bug?
11:32:51  <blathijs> (I can't easily compare with nml 0.4.5 here, because I'm building in a Debian unstable chroot where 0.4.5 no longer runs)
11:45:25  <FLHerne> blathijs: Mismatch compared to what?
11:45:53  <FLHerne> It's probably expected that the output isn't byte-for-byte identical compared to 0.4.5
11:47:56  <_dp_> this probably needs DC_EXEC check:
11:48:02  <_dp_> not the desync I looking for though
11:50:11  *** Foveafoxy has joined #openttd
11:50:52  *** Foveafoxy is now known as Guest24743
11:54:11  <FLHerne> blathijs: If it's just time.clock, making 0.4.5 run is trivial (`sed -i 's/time\.clock/time\.process_time/' nml/`)
12:09:38  *** Smedles has quit IRC
12:11:15  *** Smedles has joined #openttd
12:23:20  *** Smedles has quit IRC
12:25:07  *** Smedles has joined #openttd
12:26:57  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl opened issue #8137: New clients can't join (desync) after funding an industry
12:27:38  <_dp_> got it! xD
12:27:45  <_dp_> I'll do a pr
12:28:22  <_dp_> still feel that's not the last desync though
12:28:32  <_dp_> but it's the one I was hunting yesterday
12:30:09  <blathijs> FLHerne: IIRC OpenTTD uses md5-checksums to identify/check grf versions, so these must remain identical to the officially released versions. This is why OpenGFX release tarballs contain md5-sums of the resulting grf, which are also checked by the build:
12:34:59  <FLHerne> blathijs: That's unlikely to work across nmlc versions, then
12:35:24  <FLHerne> (even minor ones potentially, if they fix a bug that affected the output)
12:35:37  <FLHerne> Although it would be interesting to see what actually changed there
12:35:43  *** glx has joined #openttd
12:35:44  *** ChanServ sets mode: +v glx
12:35:44  * FLHerne tries
12:37:09  <Samu> CONFIRMED
12:37:33  <Samu> something to do with catchment areas
12:37:43  <blathijs> That would mean that an OpenGFX build actually has a hard dependency on the nml version used to compile it (which can be acceptable, but then the nml version should probably be included in the source tarball and I should probably update my OpenGFX Debian packages with such a hard dependency)
12:38:22  <Yexo> I don't think there is a hard requirement that opengfx builds have the same checksum
12:38:33  <Yexo> (apart from the check in the makefile)
12:39:22  <Yexo> OpenTTD indeed uses the checksum to match newgrfs, but that doesn't matter for base sets like opengfx, you can (and should) update those independent of savegames
12:39:26  *** Gustavo6046 has quit IRC
12:39:40  *** Gustavo6046 has joined #openttd
12:40:14  <Samu> let me find the PR...
12:40:46  *** Smedles has quit IRC
12:40:59  <DorpsGek_III> [OpenTTD/OpenGFX] matthijskooijman opened issue #41: Should source bundle / tarball be published on github?
12:41:09  <andythenorth> Eddi|zuHause I never clicked with F! :)
12:41:30  <blathijs> Yexo: Doesn't that mean the entire "make check" stuff can be dropped, then?
12:41:42  <Eddi|zuHause> andythenorth: that's alright, me neither :)
12:41:59  <Samu>
12:42:05  <Samu> some of these Peter PR's
12:43:01  *** Smedles has joined #openttd
12:43:06  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #8136: Add: Currency: INR
12:43:21  <Yexo> blathijs: opengfx is so simple that in general there shouldn't be many changes, which is why it worked so far
12:43:35  *** gelignite has joined #openttd
12:43:51  <Yexo> We should first understand what has changed now
12:44:25  <Yexo> See if it's a valid change or not. If valid, we can either update the "make check" stuff or remove it, not sure
12:44:47  <blathijs> That's probably a good idea anyway. I'll just stand by and wait for the experts to figure this out :-)
12:44:47  <Yexo> it is/was a safety check that nml releases don't break stuff
12:45:49  <blathijs> Hm, I thought it was more strict (e.g. like newgrf version), so the Debian package currently fails if the md5sums are different, but that might be too strict, then.
12:46:32  *** Smedles has quit IRC
12:48:16  *** Smedles has joined #openttd
12:51:13  <blathijs> (Also, something I realized recently: I *really* like the development community around OpenTTD. It's very much active, diverse and friendly. The infrastructure around automatic building, publishing binaries, etc. is very professional and project structure (commit messages, releases, changelogs, PR policies) is of a very high level. Whatever I, as Debian maintainer, throw in here gets picked up
12:51:15  <blathijs> as another chance to perfect things. All this makes it very much enjoyable to work with all you people, so many thanks for that!)
12:54:52  *** Smedles has quit IRC
12:56:54  *** Smedles has joined #openttd
12:57:57  <DorpsGek_III> [OpenTTD/OpenTTD] SamuXarick commented on issue #8137: New clients can't join (desync) after funding an industry
12:58:46  <_dp_> Samu, and even know why :p
12:59:08  <_dp_> *I
12:59:35  <Samu> where is the bug?
12:59:50  <_dp_> you'll see soon
13:00:48  <glx> missing update on industry creation, but proper rescan on load ?
13:01:45  <Samu> yes
13:02:16  <Samu> the server is wrong (and all clients that were present when the industry was created)
13:04:01  <_dp_> glx, yeah, doesn't check which tiles are actually producing
13:04:09  <_dp_> and accepting I guess as well
13:06:45  *** Smedles has quit IRC
13:08:13  *** Smedles has joined #openttd
13:11:30  *** Smedles has quit IRC
13:13:36  *** Smedles has joined #openttd
13:19:21  *** R- has joined #openttd
13:21:42  *** snail_UES_ has joined #openttd
13:23:38  *** R- has quit IRC
13:25:53  <_dp_> does industry always supply cargo on any tile?
13:26:05  <_dp_> I know acceptance is weird
13:26:47  <glx> hmm need to check
13:27:09  <_dp_> I can't even find where acceptance tiles are checked
13:27:29  <_dp_> and how it all is supposed to work when stations_near are same for all cargo types
13:29:08  <_dp_> like, I can do a simple fix for desync but this whole thing looks broken %))
13:31:07  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro closed issue #8083: Add currency Unit - Indian  Rupee
13:31:08  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro merged pull request #8136: Add: Currency: INR
13:31:11  <Yexo> Step 1: add check for broken cache in openttd.cpp along with the other desync checks
13:31:27  <Yexo> Step 2: validate it triggers on funding an industry near a station
13:31:36  <Yexo> Step 3: iterate on fix, using new check to validate if it works
13:33:31  *** matt21347 has quit IRC
13:34:08  *** nielsm has joined #openttd
13:37:26  <_dp_> Yexo, check is a separate thing, won't help with this desync
13:37:41  <_dp_> and cargo acceptance is just a weird mess, may not even desync at all
13:37:57  <FLHerne> LordAro: "ilayaraja97 and ilayarja97 committed 6 minutes ago"
13:38:04  <FLHerne> How did you manage that? ;p
13:38:34  <LordAro> ha
13:38:35  <FLHerne> (one of them's missing an 'a', too)
13:39:00  <LordAro> i noticed their commit author was broken, it's probably just taken the first 2
13:39:12  <LordAro> i assumed squash would ..squash it all together
13:44:10  <glx> I guess the issue is related to station_near
13:44:12  <FLHerne> blathijs: I built OpenGFX 0.6.0 with nmlc 0.4.5 (lightly patched so it would work) and with 0.5.1, the checksums match in both cases
13:44:22  <FLHerne> (tries with 0.5.5)
13:45:27  <blathijs> FLHerne: I haven't tried OpenGFX 0.6.0 yet (had not imported it yet), could be that there is no change there
13:49:48  <FLHerne> How do I build 0.5.5 without GIMP?
13:50:03  <FLHerne> The readme says it's optional, but `make all` fails
13:50:06  <FLHerne> (0.6.0 is ok)
13:50:26  <LordAro> comment out the relevant bits of the makefile :p
13:50:42  <blathijs> Dunno, I always build from scratch with Gimp for the .deb
13:50:44  <glx> it's optional but 0.5.5 Makefile is broken I think
13:51:06  <LordAro> alternatively, `find -name '*.png' -exec touch {} \;` might quieten it
13:51:24  <LordAro> or whatever format the images are in
13:51:29  <glx> and IIRC different gimp version can generate different PNG
13:51:35  <FLHerne> "This doesn't break the build process, but causes all PNG files to be regenerated and that causes all of them to diff from the repo."
13:51:45  <FLHerne> ^ from a GitHub issue
13:51:58  <FLHerne> Possibly that's the issue
13:52:04  <FLHerne> Actually, no
13:52:23  <FLHerne> Sprites aren't included in the grf md5
13:52:26  <FLHerne> Right?
13:53:38  <blathijs> The .deb build always regenerates using gimp, and I haven't seen an MD5-sum-failure because of that so far.
13:54:29  <blathijs> (It *could* be that the build failure I'm seeing now is really due to a different gimp version, rather than an nml version, but I have no way of telling currently)
13:55:15  <blathijs> 0;55;62m
13:56:55  <FLHerne> grfid doesn't consider the pixel data (to allow for optional 32bpp/extra-zoom versions), so I don't think that should be able to matter
13:57:08  <blathijs> Makes sense
14:00:33  *** matt21347 has joined #openttd
14:04:43  <FLHerne> blathijs: Ok, I also get a mismatch with 0.5.5/0.5.1, without doing a GIMP rebuild
14:05:08  *** supermop_Home_ has joined #openttd
14:14:06  <FLHerne> And it matches with 0.4.5
14:16:38  <Samu> FindStationsAroundTiles hmm...
14:17:02  <Samu> it has given a 3x3 rectangle of that coal mine
14:17:10  <Samu> which is wrong
14:17:38  <Samu> it should give only the tiles pertaining to the industry, only some of the tiles in that 3x3 belong to it
14:19:05  <FLHerne> Yexo, blathijs: The entire change is this:
14:19:48  <FLHerne> Old file:
14:19:50  <Samu> line 1701 industry_cmd.cpp, the ind->location should maybe be sliced into multiple 1x1 checks while iterating ind->location
14:19:55  <LordAro> something to do with the optimisations implemented?
14:19:57  <FLHerne> New file:
14:21:17  <blathijs> FLHerne: None of those numbers mean anything to me, so I can't really comment :-)
14:23:59  <glx> Samu, _dp_: I'd blame #7235
14:24:29  <DorpsGek_III> [OpenTTD/OpenTTD] ilayaraja97 commented on issue #8131: Missing bounding boxes for bridge pillars of height 1 cause graphical glitches
14:24:29  <Yexo> FLHerne: Thanks. That change is safe
14:24:48  <Yexo> It changes the offset for tram track sprites from unset (defaults to 0) to explicitly set to 0
14:25:26  <Yexo> Slight deoptimization in file size
14:25:38  <Yexo> One could argue that's a bug in nmlc, but it won't affect visible behavior
14:27:36  <Yexo> hmm, but openttd expects 119 sprites since 1.10, while apparently ogfxe_extra is still only providing 113
14:28:11  <Samu> using this->industry->location isn't totally correct
14:28:28  <Samu> i think
14:28:29  <glx> Yexo: hmm that's changed in nml 0.5.0 only IIRC
14:28:58  <FLHerne> Yexo: This is an old OGFX version, so that's fine
14:29:05  <Samu> need to look at FIRS fishing grounds
14:30:11  <FLHerne> What's the purpose of the `8n` rather than `0n` types in Action5?
14:30:34  <FLHerne> The doc says "If bit 7 is set, the offset variable needs to be set. Bit 7 may only be set if the type supports it."
14:30:40  <Samu> ah no, it's correct
14:30:48  <Yexo> nmlc explicitly sets the offset to 0 if the number of sprites is different than the maximum
14:30:53  <Samu> but let me verify for sure
14:31:24  <FLHerne> But not why I'd want to set bit 7
14:31:32  <FLHerne> Oh, because without it the offset is ignored
14:31:39  <FLHerne> But we don't need to set it anyway
14:32:01  <Yexo> FLHerne: docs are correct, but not worded very well: "1. Do not set bit 7 if the type doesn't support it. 2. Meaning of bit 7: a sprite offset follows"
14:32:51  <FLHerne> Yexo: Where's that from?
14:33:09  <Yexo> interpretation
14:33:15  <FLHerne> Oh, ok
14:33:24  <FLHerne> That makes sense
14:33:25  <Yexo> I was trying to reword it in a clearer way
14:34:34  <Yexo> Remove the "or num_sprites != num_required" in nml/actions/ line 119 to remove the diff
14:34:46  <DorpsGek_III> [OpenTTD/website] TrueBrain commented on pull request #157: Include final URL
14:36:52  <Yexo>
14:37:28  <Yexo> There we go: I added that myself apparently. Luckily I also documented the reason back then: nforenum gives a warning if the explicit offset of zero is not present
14:39:18  <Samu> Industry TileArea needs to be converted to BitmapTileArea
14:39:25  <Samu> i just don't know how to
14:39:33  <FLHerne> Yexo: Ok, makes sense
14:40:03  <FLHerne> I don't think lack of byte-for-byte identical output across major versions is a bug anyway
14:41:18  <FLHerne> blathijs will just have to patch their build script, or use OGFX 0.6.0
14:41:25  <glx> I guess nforenum will expect 113 sprites for this action 5 type
14:42:43  <planetmaker> should be ok to require a matching OpenGFX version, no?
14:43:38  <FLHerne> Well, assuming Debian has current OTTD it'll be needed anyway
14:44:01  <FLHerne> (else missing sprite warnings)
14:46:27  <blathijs> I'll probably do both, update to OGFX 0.6.0 and in that update, remove the `make check` from the build (or maybe do another upload of 0.5.5 just to remove make check maybe)
14:47:42  <_dp_> glx, well, yeah, but what to do with it now is the question
14:47:48  <_dp_> I guess Yexo's plan is ok
14:47:58  <blathijs> If everything is new (or at least the same version the original OGFX was built with), then things will work  out, but it is generally a good idea if you can also mix and match versions (and if you cannot, then version requirements should preferably be explicit, which is a bit tricky in this case)
14:48:04  <_dp_> fix desync, add check, release 1.10.2 and wait for new desyncs xD
14:49:12  <glx> I'd add the check before the fix :)
14:49:36  <glx> so it's easy to see the check works :)
14:49:56  <_dp_> glx, well, sure, do it :p
14:50:07  *** supermop_Home_ has quit IRC
14:50:19  <_dp_> I've no idea what it checks anyway
14:50:25  <glx> we're ok it's related to near_station ?
14:50:53  <_dp_> glx, yeah, I'll do a simple patch for that desync
14:51:11  <_dp_> for now at least
14:51:25  <Yexo> Isn't Station::industries_near also wrong?
14:52:05  <Samu> i found where the error is
14:52:19  <Samu> let me post a link
14:52:39  <_dp_> oh, and can someone also look at
14:52:54  <_dp_> it's a separate issue but I feel like DC_EXEC check is mising there
14:53:36  <glx> the highlightedf function is called in AfertLoad()
14:53:59  <Samu> problem here is the ind->location being 3x3, it assumes all tiles in that 3x3 belong to the industry, but only some of them do
14:54:13  <_dp_> Samu, I found it 5 hours ago :p
14:54:17  <Samu> oh, cool
14:54:22  <Samu> but i'm slow
14:54:44  <_dp_> it's not a problem to add tile check there, I'll do that right now
14:55:02  <_dp_> problem is that whole *_near stuff doesn't look right
14:55:52  <Samu> stations_near is empty at the moment
14:56:04  <glx> I think the cache check should store all industry station_near (maybe towns too), then recompute all, and compare
14:57:48  <Samu> my excelent coding skills would patch it like this
14:58:14  <Samu> just tested it, and the station no longer gets coal
15:00:51  <Samu> it's probably the wrong approach
15:09:10  <DorpsGek_III> [OpenTTD/OpenTTD] telk5093 opened issue #8138: Hotkeys don't work in non-english IME (such as Korean)
15:23:34  <Samu> _dp_ CmdRemoveFromRailStation is only called in DC_EXEC mode from what I gather
15:30:05  <_dp_> Samu, can't AI call it without?
15:30:14  *** Wormnest has joined #openttd
15:34:12  <_dp_> Samu, and actually even player can, just press shift while removing station
15:34:26  <Samu> sorry, I am wrong. What I meant to say was st->RecomputeCatchment(); isn't called in DC_EXEC mode
15:36:17  <glx> that feels wrong
15:37:26  <_dp_> whatever, I already understood it myself
15:37:36  <_dp_> there are no affected_stations without DC_EXEC
15:38:11  <glx> ok let's test the cache check I just added
15:38:18  <_dp_> would still add an explicit check just to foolproof it though
15:52:41  <Samu> DC_TEST
15:52:58  <Samu> sorry, I must be drunk
15:53:13  <Samu> it's DC_TEST
15:55:08  *** cHawk has quit IRC
15:55:11  *** Progman has joined #openttd
15:58:02  <Samu> well
15:58:25  <Samu> DC_TEST doesn't exist either
15:58:29  <Samu> im super drunk
16:05:32  <DorpsGek_III> [OpenTTD/OpenTTD] JGRennison commented on issue #8137: New clients can't join (desync) after funding an industry
16:05:38  <Samu> that feeling when I'm trying to help
16:05:46  <Samu> and end up screwing
16:07:43  <glx> ok my cache checks detects the mismatch
16:11:42  *** frosch123 has joined #openttd
16:14:05  <_dp_> lol, somehow while fixing desync I ended up with a PR that a) doesn't fix desync, b) is not needed to fix desync
16:14:12  <_dp_> probably fixes something else though xD
16:14:40  <Samu> what pr?
16:15:09  <_dp_> well, I did'n actually pr it yet
16:15:16  <_dp_> I'll just stash it for now I guess
16:15:48  <LordAro> _dp_: we're holding 1.10.2 for this, so any further fixes are welcome :p
16:16:20  <Yexo> _dp_: For desyncs such as this, the much bigger problem is finding the root cause. You've done a great job at that
16:17:19  <Samu> JGR fixes the issue in a more professional way
16:18:37  <Samu> won't make my pr then
16:20:16  <_dp_> LordAro, well, if anyone is finding newgrf house var 64 doing weird stuff I migth know why
16:20:23  <_dp_> no idea what it's supposed to do though
16:20:26  <LordAro> heh
16:21:28  <glx> JGR must have a modified CheckCaches()
16:24:03  <Yexo> He does
16:24:21  <Yexo> I can also trigger his checks without causing a desync based on some town caches
16:24:36  *** Flygon has quit IRC
16:25:43  *** tokai|noir has joined #openttd
16:25:43  *** ChanServ sets mode: +v tokai|noir
16:32:23  *** tokai has quit IRC
16:38:32  *** Guest24743 has quit IRC
16:44:13  <DorpsGek_III> [OpenTTD/OpenTTD] Andrew350 commented on issue #8131: Missing bounding boxes for bridge pillars of height 1 cause graphical glitches
16:47:07  *** luaduck has quit IRC
16:48:02  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 opened pull request #8139: Add: stations_near and industries_near cache check
16:50:52  *** luaduck has joined #openttd
17:09:12  *** Dave has joined #openttd
17:09:30  <Dave> Hello how is everyone?
17:09:51  *** Dave is now known as Guest24774
17:11:34  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl opened pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
17:12:36  <Guest24774> Trying to get world map that was on naver website? does anyone have it?
17:16:18  <Samu> I think JGR does it better
17:17:05  <Samu> now code repeats
17:21:14  <Samu> what is st->AddIndustryToDeliver(ind)
17:21:52  <Samu> gonna test anyway brb
17:22:40  <glx> yes JGR FindStationsAroundTiles() is better, but I like your AddIndustryToDeliver() use in PopulateStationNearby()
17:23:59  <Samu> but that's not how it works :(
17:24:41  <_dp_> FindStationsAroundTiles is a mess that doesn't even work for arbitrary tile areas
17:25:13  <_dp_> it's only saving grace is that it's only being called for a very specific areas
17:25:53  <_dp_> except mb for that newgrf thing that I've no idea how is supposed to work
17:27:21  <_dp_> If anything it may be better to just call RecomputeCatchment directly from PopulateStationsNearby
17:27:53  <_dp_> not as effecient but funding doesn't happen often and that leaves less opportunities for errors
17:28:06  <Samu> random spawning
17:29:02  *** JGR_ has joined #openttd
17:29:44  *** Guest24774 has quit IRC
17:30:02  *** Dave_ has joined #openttd
17:32:41  <Samu> I wonder how is AddIndustryToDeliver used, cus the station still doesn't accept coal
17:32:50  <Samu> testing with a PowerStation
17:34:44  <glx> AddIndustryToDeliver() does just what the the "/* Test if industry can accept cargo */" block did
17:36:33  *** iSoSyS has joined #openttd
17:36:35  <Samu> power station was added, but station still does not accept coal
17:37:00  <Samu> because the industry tile it reaches doesn't accept coal
17:37:16  <_dp_> yeah, acceptance is weird in openttd, but that's another story
17:37:20  <_dp_> I already talked
17:37:25  <_dp_> about it earlier
17:37:30  <glx> that's ok, you must cover the accepting tiles
17:37:46  <glx> it's known
17:37:59  <Dave_> Good evening everyone where can i get the naver world map for openttd?
17:38:00  <glx> same when you place your station near a refinery
17:38:36  <_dp_> glx, it's known behaviour but weird implementation
17:38:42  <_dp_> I also suspect some bugs
17:39:32  <Yexo> Dave_: I believe if anyone had known which map you mean you would've gotten an answer
17:39:45  <Yexo> Have you tried the in-game online content system already?
17:40:55  <Dave_> the world map with real city locations and names.
17:42:45  <Yexo> Have you tried ?
17:43:41  <FLHerne> Dave_: There's more than one of those ;-)
17:44:55  <Dave_> thanks
17:44:56  <glx> hmm I should probably adapt for #8139, seems cleaner
17:45:17  <Yexo> Dave_: Probably you can find it on, but I can't find it using a translator
17:45:37  <Dave_> tried it not the one i am looking for thanks
17:45:47  <DorpsGek_III> [OpenTTD/OpenTTD] DorpsGek pushed 1 commits to master
17:45:47  <DorpsGek_III>   - Update: Translations from eints (by translators)
17:45:59  <Dave_> thanks will try
17:47:14  <Samu> ok, I see
17:47:16  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on pull request #8139: Add: stations_near and industries_near cache check
17:48:11  <JGR_> I can turn that into a PR if useful?
17:48:22  <JGR_> In general I've got loads of desync-related stuff in my branch
17:48:44  <glx> I'll adapt it myself, lots of code changed in this area anyway
17:48:57  <glx> all the loops, ...
17:49:43  <JGR_> It's probably be simpler to copy from a more recent version, the loop changes have been done already
17:49:50  <JGR_> It'd*
17:51:36  <LordAro> JGR_: i'm sure any desync stuff would be appreciated :)
17:51:39  <LordAro> clearly we need it
17:52:13  *** cHawk has joined #openttd
17:53:30  <_dp_> speaking of more desync stuff... one more random ugly hackish tool to compare saves fwiw:
17:54:58  *** Dave_ has quit IRC
17:55:36  <JGR_> I've been getting desync bug reports from users for a while now
17:56:11  <JGR_> Reproducing them from user feedback tends to be difficult, so I've tried to automate things a bit
17:56:27  <glx> indeed you check a lot more than us, that clearly could help
17:58:00  <JGR_> In my branch, when a desync does occur, CheckCaches is run on the client and server, any output from these is added to a log resembling a crashlog
17:58:15  <JGR_> The client and server exchange logs and these are written to files, and into savegames
17:59:46  <JGR_> As well as using the random seed to detect desyncs, I'm also hashing key bits of game state into a different seed which is compared between client and server
18:00:51  *** mcbanhas has joined #openttd
18:00:56  <mcbanhas> Hello
18:01:03  <_dp_> JGR_, but can't server already be ahead when it receives desync report from the client?
18:01:32  <JGR_> Yes, so the savegames would not be identical
18:01:53  <_dp_> that doesn't seem to be very helpful then :(
18:02:35  <_dp_> 1 tick of simulation is already enough to mess up the whole map
18:02:37  <JGR_> However running CheckCaches immediately often finds the problem, and interpreting the output of that requires a savegame
18:09:28  <JGR_> To give an example, is what led to #7856
18:10:27  <LordAro> JGR_: if you can add some stuff in the doc at the same time, would be much appreciated
18:11:21  *** Wolf01 has joined #openttd
18:15:18  *** supermop_Home has joined #openttd
18:15:37  <supermop_Home> hi
18:15:50  <Wolf01> o/
18:16:14  <supermop_Home> two neat new grfs bringing me join
18:18:41  <supermop_Home> joy
18:18:50  <supermop_Home> man my brain is broken
18:18:53  <JGR_> LordAro: Sure, can do.
18:20:06  <JGR_> Does all of the stuff I outlined sound reasonable, or just a subset? I'm sure there's a line somewhere of changes which are too invasive.
18:20:53  <supermop_Home> Opengfx+ stations, and town layout
18:24:44  <LordAro> JGR_: seems reasonable to me
18:25:11  *** plstc has joined #openttd
18:25:38  <DorpsGek_III> [OpenTTD/OpenTTD] JGRennison opened pull request #8141: Fix #8137: Rectangular industry area used for stations_near calculation
18:30:29  <_dp_> great, so now we have competing prs :p
18:30:32  <FLHerne> frosch123: Is SpriteCache used just for caching files between nmlc runs, or also to deduplicate sprites or to avoid re-processing them during a single run?
18:30:46  *** plastic has quit IRC
18:31:03  <FLHerne> (i.e. if I stopped using it altogether when --no-cache is passed, would that break stuff?)
18:31:33  <_dp_> imo if change FindStationsAroundTiles it should be made into template and use_nearby moved to StationFinder
18:33:33  *** plastic has joined #openttd
18:33:48  <JGR_> _dp_: Happy for you to take precedence on this
18:34:46  <JGR_> Mine can be closed if something else is merged
18:39:26  *** plstc has quit IRC
18:39:26  *** plastic has quit IRC
18:40:39  *** plastic has joined #openttd
18:41:37  <_dp_> it's just funny that of all the bugs to work on this stupid thing that no one cared about for year suddenly gets 3 patches in one day xD
18:42:31  <frosch123> FLHerne: 1 and 2, not 3
18:44:53  <frosch123> FLHerne: disabling it probably means to keep all sprites in memory, so in case of 32bpp grfs some 100 mb more address space
18:45:17  *** plstc has joined #openttd
18:48:52  <mcbanhas> Hi guys, would anyone be kind enough to give me a hand with git? I'm ready to submit my first batch of string changes, and first and foremost I'm not sure whether I should create a new PR or edit the old one.
18:49:03  <frosch123> FLHerne: actually, it also somewhat does 3. there was a big performance gain by opening huge source images only once, instead of repeatedly for every single sprite in them
18:50:20  <frosch123> FLHerne: why do you not want to use the cache? i mean you can just delete it, if you do not want it
18:50:35  <frosch123> wasn't there an option to hard-reset it?
18:51:16  *** plastic has quit IRC
18:51:46  <frosch123> lol, "nml --help" says that disabling the cache may speed up :p
18:52:08  <frosch123> i think you have to hand-craft that scenario
18:56:09  *** JGR_ has quit IRC
18:57:18  <Samu> how does this checkcache starts? is there a command?
18:59:33  <Yexo> It's executed when debug_level_desync >= 1
19:00:25  <Samu> a
19:00:40  <Samu> 2 suffice?
19:00:57  <Samu> i typed 9 and this is writting savegames somewhere
19:01:22  <Yexo> >=1, so 2 is fine, as is 1
19:01:44  <LordAro> Samu: i think you know that 2 >= 1 by now
19:02:53  *** Gustavo6046 is now known as Guest24787
19:02:57  *** Gustavo6046 has joined #openttd
19:02:59  <_dp_> @calc 2>=1
19:03:00  <DorpsGek> _dp_: 1
19:03:51  *** Guest24787 has quit IRC
19:06:43  <Samu> oh, right, i need 8139
19:06:55  <Samu> that's why i wasn't seeing anything
19:12:46  *** frosch123 has quit IRC
19:14:34  <glx> and it won't print on the console, it's only in commands-out.log
19:15:10  *** frosch123 has joined #openttd
19:16:37  <glx> <@Yexo> It's executed when debug_level_desync >= 1 <-- it's skip for <=1, so 2 is the minimim :)
19:17:06  <Yexo> Thanks for the correction :)
19:23:52  <FLHerne> frosch123: There's a bug where it creates the cache directory even if --no-cache is set
19:24:20  <FLHerne> The most straightforward fix would just be to not create the cache if it wasn't really being used
19:24:53  <FLHerne> But since it is, I'll do something else
19:25:12  <Samu> glx, it corrects the cache, is it supposed to do that?
19:25:39  <FLHerne> And yes, I saw that description, and couldn't figure out if it's supposed to say "disable (the cache, which might be faster)" or "(disable the cache), which might be faster" :P
19:25:55  <glx> yes as all cache checks it fixes the cache while checking
19:26:06  <frosch123> FLHerne: it's probably an ancient message
19:27:00  <glx> but it still reports the error it any
19:27:04  <frosch123> nml < 0.4 only dumped sprites into the cache. only with >= 0.4 there were significant gains with processing sprites out of order
19:27:45  <Samu> ok
19:28:08  <Samu> makes it harder to get clients desync this way
19:28:31  <Samu> seems counterproductive
19:29:58  <FLHerne> frosch123: ok
20:16:30  <glx> Samu: cache check is there to detect possible desyncs, and it also prevents them, but it's cpu intensive, so only enabled when hunting for desync cause
20:35:33  *** plstc has quit IRC
20:35:52  *** plstc has joined #openttd
20:42:58  <LordAro> glx: so what are we doing about the cache check?
20:44:48  <LordAro> other than also checking catchment_tiles, i don't really see how JGR's solution is "cleaner"
20:45:18  <glx> mine is using a map, vector is better
20:47:06  <LordAro> eh, marginally
20:49:54  <andythenorth> ha ha, I am going to hate drawing cargo icons for all these chemicals
20:50:11  <andythenorth> methanol, ethylene, propene etc
20:50:20  <LordAro> glx: well, feel free to rewrite yours to use vectors, i guess
20:50:29  <andythenorth> maybe I can fit the chemical symbol in an icon :P
20:50:32  <LordAro> due to the FOR_ALL removal, it'd have to be basically rewritten anyway
20:50:39  <andythenorth> or the bonds?
20:50:59  <andythenorth>
20:51:27  <glx> LordAro: yeah I'll switch to vectors (watching twitch for now)
20:51:59  <LordAro> glx: any particular thoughts about the solution?
20:52:09  <LordAro> either of the solutions*
20:52:23  *** Wormnest has quit IRC
20:53:35  <glx> I'd say a mix of both, I prefer JGR solution, but I like the AddIndustryToDeliver() usage
20:53:43  <LordAro> mm
20:54:27  <LordAro> tbh i'd suggest that FindStationsAroundTiles should be an Industry member function
20:54:43  <LordAro> actually, no
20:54:52  <glx> it's used for towns too I think
20:54:57  <LordAro> yeah
20:56:15  <glx> but it should maybe merge into StationFinder
20:56:22  <DorpsGek_III> [OpenTTD/nml] jrook1445 opened issue #140: 0.5.1 - animated forest tiles from OpenGFX+ Industries disappear
20:56:48  <_dp_> use_nearby should go to stationfinder as it doesn't work without anywaya
20:57:31  <_dp_> and mb separate finder for houses
20:58:00  <_dp_> use_nearby for industry can be dropped completely, it's not used anyway
20:58:23  <LordAro> mm, there's a certain amount of "trying to be too generic" going on
20:58:46  <glx> templated StationFinder() maybe
20:59:13  <DorpsGek_III> [OpenTTD/nml] FLHerne commented on issue #140: 0.5.1 - animated forest tiles from OpenGFX+ Industries disappear
20:59:21  <LordAro> unlikely to be worth it, i'd think?
20:59:50  <_dp_> finder is already an iterator so no point templating
21:00:00  <_dp_> FindStationsAroundTiles can be templated though
21:00:14  <_dp_> in my pr I already kinda prepared for templated FindStationsAroundTiles
21:00:34  <LordAro> _dp_: perhaps you should finish it :)
21:00:51  <_dp_> well, I guess
21:01:02  <_dp_> I already have most of it stashed anyway :)
21:01:12  <_dp_> should it be same pr though?
21:01:17  <_dp_> has nothing to do with desyncn
21:03:00  <glx> anyway I'm not fan of copy/pasting FindStationsAroundTile() inside PopulateStationsNearby()
21:04:23  <LordAro> _dp_: possibly not, but if it's part of the same work, there's no good arbitrarily separating it
21:05:34  <glx> I'm not against a multi commit PR with first rewrite to ease fixing, then the fix itself
21:05:39  <_dp_> it depends on one another, I either copy FindStationsAroundTile or template it
21:06:40  <_dp_> multicommit prs are a pain to change if needed
21:07:07  <_dp_> or at least I've no idea how xD
21:07:20  <glx> rebase magic
21:07:42  <LordAro> git commit -m "tmp" && git rebase -i with fixup into the original commit
21:07:49  <LordAro> fix any conflicts if necessary
21:07:52  <michi_cc> _dp_:
21:07:54  <LordAro> that's how i do it, anyway
21:08:14  <michi_cc> Though sacrilegious for some, a GUI tool can help tremendously with that :)
21:08:46  <glx> and to split a commit you put it twice in the rebase with "edit" on the first one
21:09:17  * _dp_ never liked git
21:12:40  *** tokai has joined #openttd
21:12:40  *** ChanServ sets mode: +v tokai
21:13:50  *** Progman has quit IRC
21:17:02  *** iSoSyS has quit IRC
21:19:33  *** tokai|noir has quit IRC
21:22:21  *** andythenorth has quit IRC
21:24:15  *** Samu has quit IRC
21:25:25  <mcbanhas> >me is trying the vactrain newgrf for the first time
21:25:34  <mcbanhas> oh god this is amazing
21:31:49  <FLHerne> Trains, but really fast?
21:32:07  *** frosch123 has quit IRC
21:38:46  <mcbanhas> and in tubes
21:39:27  <FLHerne> Vacuuuuuum tubes, no less
21:40:40  <mcbanhas> it's sad, because we wont be seeing stuff like that IRL during our lifetimes
21:41:02  <mcbanhas> Even most maglev projects sound like pie in the sky at the moment.
21:42:14  *** Wormnest has joined #openttd
21:44:10  <_dp_> I hate C++ templates
21:44:20  <_dp_> now every stupid thing wants to go to h file :(
21:44:49  <LordAro> :/
21:48:34  *** cHawk- has joined #openttd
21:49:15  <_dp_> what's next junk h name after _map.h and _func.h ?
21:52:01  <nielsm> there's _type.h but that's mostly basic typedefs and enums
21:52:06  <_dp_> it should rly got to _func but _map depends on _func and I can't untangle that mess :(
21:52:27  <nielsm> otherwise invent a new one
21:52:39  <Yexo> Which headers are that?
21:52:47  <_dp_> station_junk.h
21:52:58  <nielsm> if you can avoid including a large chunk of code everywhere that will help compilation times a bit
21:54:41  *** cHawk has quit IRC
21:55:15  <glx> we use .hpp for templated drivers
21:55:50  *** Wolf01 has quit IRC
21:57:43  <glx> hmm no they don't use templates
21:58:30  *** sla_ro|master has quit IRC
22:05:06  *** nielsm has quit IRC
22:06:57  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 updated pull request #8139: Add: stations_near and industries_near cache check
22:22:15  *** matt21347 has quit IRC
22:25:58  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro approved pull request #8139: Add: stations_near and industries_near cache check
22:41:17  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl updated pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
22:41:35  <_dp_> didn't quite work out as planned but still ok I guess
22:42:24  <_dp_> bloody tabs
22:44:58  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl updated pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
22:51:09  <LordAro> yeah, commits aren't exactly separated properly
22:53:13  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
22:55:34  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
22:56:24  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
22:58:25  <_dp_> LordAro, templated lambda = inline
22:58:46  <LordAro> last i heard they were often inefficient
22:59:22  <_dp_> if called as function pointer mb but as templated that's the whole point
23:00:26  <_dp_> unlike vector that requre some lto magic
23:00:38  <DorpsGek_III> [OpenTTD/OpenTTD] LordAro requested changes for pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:00:41  <michi_cc> Any compiler/optimizer that is not totally dumb should realize these lambdas as a simple stack-stored struct with a single member var and an operator(). Which the optimizer can possibly inline away.
23:00:43  <LordAro> possibly
23:01:53  <LordAro> bedtime
23:02:01  <LordAro> do have a go at splitting those commits up a bit better
23:02:08  <LordAro> they're a mess at the moment
23:02:44  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:03:00  <_dp_> LordAro, if you mean extra TileIndex it's only needed for desync fix
23:03:09  <_dp_> completely unnecessary without
23:04:44  <glx> I would move AddIndustryToDeliver() change to its own commit
23:05:06  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:06:26  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:07:08  <glx> (I know it's not so easy to split a commit, but in this case it should be doable
23:10:46  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:13:14  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:14:20  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl updated pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:17:52  *** mcbanhas has quit IRC
23:19:54  <DorpsGek_III> [OpenTTD/OpenTTD] glx22 merged pull request #8139: Add: stations_near and industries_near cache check
23:20:28  *** Yexo has quit IRC
23:21:42  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl updated pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:23:11  *** Yexo has joined #openttd
23:23:11  *** ChanServ sets mode: +o Yexo
23:24:11  <DorpsGek_III> [OpenTTD/OpenTTD] ldpl commented on pull request #8140: Fix #8137: New clients can't join (desync) after funding an industry
23:32:49  *** Wormnest has quit IRC

Powered by YARRSTE version: svn-trunk