Time Nick Message 00:25 VanessaE LazyJ says that he cant see me on the server. He says when I move there is only a blackoutline 01:31 kaeza VanessaE, sorry for the delay 01:31 VanessaE s'ok 01:31 kaeza any more info needed? 01:31 VanessaE for #863? I suppose not 05:14 celeron55 02:00:38 < pitriss> Please can any MT dev look here https://forum.minetest.net/viewtopic.php?id=6197&p=12 -- 05:14 celeron55 the server now says that when it gets in the "failed to emerge player" state 05:15 VanessaE celeron55: someone on my server also keeps getting that error, even when the server's almost totally idle. 05:15 VanessaE (probably the same reason you just stated, though, hard to say as he didn't give me much info) 05:16 celeron55 get some relevant log of it through multiple login attempts 05:19 VanessaE *looks at log and groans* 05:19 celeron55 i was not aware that people would consider it better for a server to say nothing when it can't allocate the player than say that the client can't get it right now 05:19 VanessaE any particular search pattern that'll get what you need? this is ather verbose 05:19 VanessaE rather* 05:19 celeron55 maybe they're just used to waiting for the timeout when connecting 05:19 celeron55 if their client crashed 05:20 VanessaE it might just be the choice of wording that's the issue. 05:20 celeron55 well, i'd like to know if that happens in any other case than when the client's connection has broken right before it (which shows up as a connection timeout occurring sometime after) 05:21 VanessaE maybe write something like "The server says you're still online. Maybe your client exited unexpectedly? Wait {some amount of time here} and try again." 05:23 VanessaE hm, lots of 'New connection: "his user ID here" from his_IP_here', without disconnect/timeout messages between them 05:24 VanessaE http://pastebin.ubuntu.com/5953903/ 05:24 VanessaE e.g. ^^^ 05:24 celeron55 the "access denied" part of the message is generated by the client 05:24 VanessaE (note the timestamps...every 2 seconds seems like...a problem) 05:25 VanessaE (line 2 is from the player_textures mod) 05:26 celeron55 hmm... i'll add a few more loggings from the server; it doesn't log certain relevant situations here 05:26 VanessaE a minute later I see stuff like this: http://pastebin.ubuntu.com/5953907/ 05:27 VanessaE (obviously I just grep'd the log for the userid) 05:29 celeron55 hmm actually... i added the relevant log messages already 05:29 celeron55 oh 05:29 celeron55 add -A 3 or so to grep 05:29 VanessaE sure 05:29 celeron55 it doesn't print the player name in all messages there 05:30 VanessaE a bit more verbose (with the aforementioned -A 3): http://pastebin.ubuntu.com/5953914/ 05:39 VanessaE (I tend to forget about grep's context switches) 05:59 celeron55 i think i'll make it say in that situation "An another client is connected with this name. If your client closed unexpectedly, try again in a minute." 06:01 VanessaE that's good 06:01 VanessaE s/An another/Another/ 06:05 celeron55 http://paste.dy.fi/lJX 06:06 celeron55 also what's up with the message dialog 06:06 VanessaE ? 06:07 celeron55 the bottommost line gets cut off in a way that makes you wonder if there's more text in there than it's showing 06:07 VanessaE oh yes 06:07 celeron55 probably a bug in the TTF code, altough it might be older than that 06:07 VanessaE always been cut off like that for...ever 06:07 celeron55 it is? well, i'll fix that too 06:08 celeron55 the dialog is very horrible anyway though 06:08 celeron55 irrlicht's word wrapping is too brainless for error messages coming from lua and so 06:08 VanessaE heh 06:09 celeron55 maybe i'll leave that for some time when i'm not sitting on my computer after breakfast instead of going to work 06:09 celeron55 hmm... i mean, at my computer 06:10 celeron55 sitting on it would be even more useless though 06:10 VanessaE well it gets kinda cold up there in .fi, so "on" might be a useful way to keep your butt warm :P 06:10 celeron55 my butt has a built-in heater! 06:11 VanessaE gas-powered I'm guessing ;) 06:23 VanessaE celeron55: anyhow, line 108, s/An another/Another/ 06:23 VanessaE else I guess it's right, you know the code better than me :) 07:29 SpankDat hi 07:29 SpankDat :D 07:29 * SpankDat all on the floor 07:29 * SpankDat give me some more 07:29 * SpankDat 'til you get sore 07:30 * SpankDat , ooohh.. 07:31 VanessaE ... 07:43 sfan5 I'm totally going to ban him from #minetest 07:45 sfan5 s/totally/definitly/ 07:46 sfan5 s/definitly/definitely/ 08:44 VanessaE sfan5: I thought he already was 10:36 sfan5 VanessaE: he was, but changed his nick 10:39 Jordach you know the texture pack selection pull; i can switch between VanessaE's HDX and Default without issues 10:41 Jordach it even works with preload item visuals 11:23 nore can someone merge #860? 11:24 nore hmmmm said he was OK, and kahrl said that if hmmmm was, then he was too 11:24 nore (logs from yesterday) 11:24 PilzAdam sure 11:24 Jordach nore, your texture pack selector works: i can switch between HDX and Default WITHOUT issues 11:24 nore Jordach, thanks 11:24 Jordach in one ./minetest 11:24 nore the most difficult was to clear the texture_to_path cache 11:25 Jordach i dont have to keep killing minetest 11:25 nore because you would else have to restart 11:25 Jordach it even works with preload item visuals 11:25 nore PilzAdam, did you look at #861? 11:25 nore Jordach: I'm happy it does 11:29 PilzAdam nore, 861 works 11:29 nore thanks 11:41 nore PilzAdam, why isn't #856 merged? 11:48 nore what of you think of #688 too? 13:03 PilzAdam a simple fix for worlds in the server tab being filtered by selected game in singleplayer tab: https://gist.github.com/PilzAdam/6164236 13:08 iqualfragile thats a funny bug 14:52 celeron55 http://paste.dy.fi/lXn <- i will push this in a moment ("Clean up server's log messages and give a better error to client when it's player is in use") 14:52 celeron55 oh, the commit message is bad; it should be "its" 8) 15:00 PilzAdam celeron55, wait 15:01 PilzAdam BlockMen said that MSVC complains when you do L"foo"\n"bar", it expects L"foo"\nL"bar" 15:02 BlockMen PA, then merge #859 before celeron55 commits 15:05 celeron55 wud 15:06 celeron55 what the 15:06 celeron55 don't merge that 15:06 celeron55 what kind of bullshit is that LagPool modification for example 15:06 celeron55 there should be absolutely no reason for that whatsoever 15:07 BlockMen when compiling with MSVC it takes #define max instead the float value 15:07 BlockMen i have no idea why, but it dies 15:07 celeron55 eh 15:07 BlockMen *-i +o 15:08 celeron55 1) that is ridiculous 15:08 celeron55 2) the fix is ridiculous 15:08 celeron55 change the name of pool and max to m_pool and m_max or something sane like that 15:08 PilzAdam or s/#define max/#define MAX 15:09 BlockMen ok, then just close the request 15:09 celeron55 PilzAdam: it's in MSVC's headers; one'd need to #ifdef MSVC #undef max #endif or something like that 15:09 celeron55 and then you'd stumble upon different versions of MSVC doing it differently 15:10 PilzAdam that is ridiculous 15:10 celeron55 anyway, i'll take care of those 15:11 celeron55 requiring L in the middle of strings is also ridiculous 15:12 celeron55 maybe we should make a list of things in MSVC that are ridiculous and publish it, just for lulz 15:15 BlockMen celeron55, btw thanks for calling my fix atempt (where PA helped me btw) ridiculous 15:16 celeron55 BlockMen: don't take it personally 15:17 celeron55 thinking you are code is silly 15:18 celeron55 (anyway, good catch as always) 15:19 celeron55 pushed the server message stuff and the better MSVC fix for lagpool 15:23 PilzAdam what about that mainmenu fix i posted earlier? 15:24 BlockMen celeron55, ok. 15:26 celeron55 PilzAdam: it seems reasonable 15:30 celeron55 PilzAdam: of course there could be a better system for setting what linked stuff should be shown with each tab though 15:30 PilzAdam Im not very familiar with sapiers code 15:31 celeron55 i wouldn't attempt that 15:38 BlockMen celeron55, hmmmm and other devs, what you say to #851 ? 15:42 celeron55 ha 15:42 celeron55 this broken text formatting in the error message dialog is sapier's lua mainmenu's fault 15:43 celeron55 i wonder how much time has already been wasted fixing all this stuff up again 15:43 kahrl BlockMen: seems good, are there compatibility issues? 15:43 hmmmm 851 looks good 15:43 kahrl oh, I think you need to add drowning to minetest.nodedef_default in builtin/item.lua 15:44 BlockMen kahrl, im kinda sure that there are not, but let me check again 15:44 PilzAdam Ill merge 851 then 15:45 kahrl I mean how many mods make use of the old drowning system and will they get updated 15:45 hmmmm oh you know, when i read the original code i thought that tried to read it as a bool first then an int 15:45 kahrl I think they will since it's a kinda new system 15:45 PilzAdam kahrl, previously all liquids drowned the player 15:45 PilzAdam now you need to add it to nodedef 15:46 kahrl hrm 15:46 proller hmmmm, maybe decrease nparams_biome_def_humidity 50->20 before release ? 15:46 PilzAdam but we never released the old system 15:46 hmmmm then this is fine 15:46 kahrl so all mods that register liquids are affected, even very old ones 15:46 hmmmm hey proller 15:46 kahrl that predate the drowning 15:46 PilzAdam kahrl, if you compare 0.4.7 -> 0.4.8 nothing changes for them 15:46 hmmmm i was just wondering, what is the purpose of sending the heat and humidity to the client? 15:47 proller hmmmm, for making fog later 15:47 kahrl I guess you can see it that way, and I agree it's probably no big deal 15:48 hmmmm proller, is that actually needed? 15:48 celeron55 you definitely need to add the field to minetest.nodedef_default 15:48 hmmmm what if there was a different packet to add extra data to mapblocks or specific positions or areas 15:48 BlockMen ok, will add then 15:48 PilzAdam BlockMen, Ill do that 15:48 BlockMen PA, y? 15:49 PilzAdam I already have it squashed and merge ready 15:49 celeron55 (all fields should be there so that mods can check them without checking for nil or other garbage) 15:49 BlockMen PA, ok 15:50 proller hmmmm, actually its useful for debug 15:51 proller hmmmm, some mods now can start using heat/humidity, and debug info about it will helpful 15:52 proller about heat/humidity at your position 15:53 kahrl proller: you could make a craftitem that prints the heat/humidity at the player's position in on_use 15:53 kahrl it's only needed on the server that way 15:53 BlockMen PilzAdam, and https://github.com/minetest/minetest_game/pull/194 for updating drowning in _gmae 15:53 BlockMen *_game 15:53 PilzAdam already done 15:54 proller kahrl, yes, but i want make fog depended on humidity, and its imposible without sending to client 15:54 BlockMen good 15:55 PilzAdam kahrl, btw, found a bug in your last crack overlay commit: the last frame of the crack is shown if you release the dig button 15:57 proller and now possible to make mod with temperature damage, and maybe make gui for heat at client with it 15:57 kahrl oh dang, how did I not notice that 15:58 proller hmmmm, and what about decrease nparams_biome_def_humidity 50->20 before release ? 16:00 proller fog better to make after https://github.com/minetest/minetest/pull/799/files 16:01 PilzAdam I dont know if 799 gets merged 16:02 BlockMen a short question: would a hand like that get merged? -> https://dl.dropboxusercontent.com/s/1sbvb4xeeerd0tg/screenshot_356446763.png 16:02 BlockMen or do you guys think its too mc like? 16:04 celeron55 proller: by the way, it's likely that the weather data transferred over network is yet moved to an another packet; don't be too hasty about moving forward with what it is 16:04 celeron55 proller: but it likely requires hmmmm's attention and time, the availability of which i'm not really knowledgeable of 16:06 kahrl PilzAdam: the problem is that I changed progression to a u32 16:06 kahrl then progression >= 0 is always true 16:08 proller celeron55, i start playing with improving clouds, and sending clouds_height from server, it maybe needs in new packet, and maybe make yet another packet for mapblock near it 16:08 proller what else send for mapblock? 16:09 celeron55 the cloud thing is also quite controversial 16:09 celeron55 don't expect to get them in without talks with everybody; it's just so different to what we have currently 16:09 kahrl PilzAdam: this fixes it: https://gist.github.com/kahrl/6165933 16:11 proller celeron55, now its small try, if it will improvement - i will make pull with discussion and "NO" from PilzAdam ;-) 16:12 PilzAdam kahrl, yep, that works 16:14 kahrl shall I push it? 16:14 PilzAdam sure 16:14 hmmmm proller, at this point, there are a lot of things that need to be changed/completed/fixed before 0.4.8 comes out 16:14 hmmmm i don't think we're going to see it for a while 16:14 hmmmm but yes, that will be changed before release, definitely 16:15 kahrl hmmmm: that is going to make RBA sad ;) 16:15 hmmmm it's because everybody started adding these huge features that half work 16:15 PilzAdam what about the 0.4.8 feature freeze? 16:15 hmmmm I suppose there could be an indefinite feature freeze 16:15 hmmmm at this point we don't need more new stuff added, the crap that's already here needs to be fixed and there's a lot of it 16:16 kahrl would a feature freeze encompass the httpfetch framework I'm working on? 16:16 hmmmm hmm 16:16 PilzAdam I think so 16:16 hmmmm i'd say no because there's not much that could interfere 16:17 PilzAdam kahrl, how long do you need to finish it? 16:17 kahrl it could break things because I'm planning to change every place that uses curl 16:17 hmmmm oh ;/ 16:17 kahrl PilzAdam: dunno, depends how difficult it is to integrate with the main menu 16:17 kahrl and how the speed tests that I'm going to do work out 16:18 PilzAdam I think it should wait 16:18 kahrl but we can't release 0.4.8 with these huge freezes in the mainmenu 16:18 kahrl when it can't connect to some server 16:19 celeron55 it also looks like we're not going to get polish translations if it's to be included that magazine 16:19 celeron55 +with 16:20 celeron55 so it's kind of useless to try to aim for that 16:20 PilzAdam sapier seems to hide ;-) 16:20 kahrl when RBA came here to announce the magazine thing, I thought to myself that it's not going to happen 16:20 kahrl there just wasn't enough time 16:22 PilzAdam https://github.com/minetest/minetest/pull/861 this seems to be a simple fix 16:22 PilzAdam s/fix/improvement/ 16:23 Calinou https://github.com/minetest/minetest/issues/864 \o/ a local "standards guy" 16:25 kahrl PilzAdam: what is the reasoning behind the current complicated comparison function? 16:26 kahrl is there some corner case it tries to address 16:26 PilzAdam I wrote it once since I didnt knew that you can compare strings like that; and sapier just copied it from me 16:26 kahrl ah 16:26 BlockMen btw issues -> https://github.com/minetest/minetest/issues/763 16:27 BlockMen im using it in my builds and it seems to be fine 16:27 kahrl PilzAdam: does the new version sort "same" strings in an undefined order? 16:27 kahrl "same" as in "World" vs. "world" 16:28 PilzAdam yes 16:29 kahrl perhaps if a.name:lower() == b.name:lower() then return a.name 0.4.8? 18:54 celeron55 hmm, has none of this been in 0.4.7? 18:54 BlockMen drowning was just enabled for dev builds 18:56 BlockMen celeron55, it was added after 0.4.7 release -> https://github.com/minetest/minetest/commit/53066024f6a91d5f83241b379b94d8557d43a646 18:56 celeron55 and actually, that doesn't even matter 18:56 celeron55 it'd be relatively new anyway 18:56 celeron55 okay, it's fine to me as-is (assuming there isn't some error i didn't notice) 18:58 BlockMen at least 3 more devs have checked, so there shouldnt be 19:01 VanessaE bbl 19:44 kahrl hey, I found a memleak in the server_announce code 19:44 kahrl the string allocated by curl_easy_escape is not freed 19:45 kahrl but I'm rewriting it anyway :P 19:45 celeron55 oh god, the formspec field is just bad; it appears to be impossible to make it show a multiline error message properly 19:45 celeron55 the error_message is shown using it now and... eh 19:45 celeron55 it sucks 19:46 celeron55 if you get anything that doesn't fit on a single line (like you get half of the time), you don't see some pixels of the second line and none of the lines after that 19:46 celeron55 should this be somehow hacked together or does anyone have any interest in maybe making it properly 19:47 celeron55 making the field larger doesn't make more text fit to it 19:48 kahrl celeron55: what would be involved in making a proper solution? 19:48 celeron55 fixing something in guiFormSpecMenu maybe 19:49 celeron55 see builtin/mainmenu.lua:147 (or somewhere around there) 19:50 kahrl perhaps use a textarea instead of a field? 19:50 kahrl a readonly one if that is possible 19:50 celeron55 well, i'll try 19:50 celeron55 readonly isn't possible 8) 19:50 kahrl that would also make it possible to copy the message to the clipboard (if clipboard works in irrlicht) 19:54 VanessaE back 19:57 celeron55 the formspec textarea's IGUIEditBox doesn't have word wrapping turned on 19:57 celeron55 is there some application where it shouldn't be turned on? 19:57 celeron55 the word wrapping is very dumb though; it just can't wrap a filesystem path for example 19:57 VanessaE ok, signing onto servers when there's been any kind of recent timeout is completely busted now :-/ 19:58 kahrl celeron55: some stuff like luacontrollers might be newline sensitive (I don't use them so I don't know) 20:00 celeron55 this is how it looks like when using a current textarea http://i.imgur.com/3C7hl6U.png 20:00 celeron55 (it doesn't look much better when adding word wrap) 20:00 kahrl well yuck 20:01 celeron55 8D 20:01 celeron55 the formspec mainmenu is pretty horrible 20:02 celeron55 that's with irrlicht 1.7 though; with 1.8 the background color would be green 20:04 kahrl whenever there are formspec problems like this my temptation would be to add special-purpose formspec element types 20:04 kahrl (like a errormessage[...] field in this case) 20:04 kahrl not the greatest idea ever :P 20:05 PilzAdam the only thing that works good in formspecs are lists 20:05 PilzAdam everything else just feels wrong 20:06 PilzAdam you can never get anything pixel perfect at every window size 20:06 celeron55 well it wasn't designed for anything it is used for now; it was designed for just showing inventories 20:06 PilzAdam why do we use it for the menu then? 20:06 celeron55 i have no idea 20:06 celeron55 ask sapier 20:07 celeron55 or whoever thought it was a good idea 20:07 celeron55 i literally just came back once and saw that it had been added 20:08 celeron55 or, well, maybe someone asked something but i wasn't interested 20:16 celeron55 it's not like the alternative is that glorious either though 20:18 celeron55 but these aren't really tolerable 20:33 proller update jsoncpp - https://github.com/proller/minetest/commit/ea346f43629c90301267176f2e0b1cde6231d18f 20:36 VanessaE ok we got three servers now with the "could not allocate player" issue - mine, Landrush, and some "Remobo" server (whichever that is) 20:36 VanessaE aside from mine being more up-to-date and showing the newer message. 20:38 VanessaE (Landrush's issue description starts here, https://forum.minetest.net/viewtopic.php?id=6197&p=12 ) 20:40 celeron55 something obviously is wrong in some unforeseen way 20:42 celeron55 VanessaE: can you try commenting out the DenyAccess() on lines 2059...2061 on your server 20:42 celeron55 and see what happens 20:43 celeron55 (server.cpp) 20:44 VanessaE sure, sec. 20:46 VanessaE wait, I haven't restarted it yet :) 20:47 celeron55 i want to try before that 20:47 VanessaE oh ok 20:47 VanessaE timed out already 20:47 VanessaE (that was quick) 20:47 celeron55 i failed a password and now i'm getting "it" 20:47 celeron55 now i tried again after a short pause and now it works 20:48 VanessaE timed out. 20:50 VanessaE and again. 20:50 VanessaE and twoelk just signed out - he's one who was having problems before. 20:52 celeron55 this doesn't make any sense 20:54 celeron55 oh well, try something 20:54 VanessaE hm? 20:55 celeron55 if the DenyAccess comment-out doesn't affect it (it'll probably just leave users out without notifying them at all); you could try disabling the if block at server.cpp:4653 20:55 celeron55 i'm going to sleep now 20:56 celeron55 i guess this is to some extent reproducible on a random test server too so i'm not sure if it's worth it to test it on your main server 20:57 VanessaE actually, you'll have a hard time reproducing this on a local server - I *never* have this error. 20:57 VanessaE I suspect at least some routine network latency is necessary for it to occue 20:57 VanessaE occur* 20:58 VanessaE I'll leave out the suggested changes for now. 22:57 kahrl valgrind is reporting an uninitialized access in formspec 22:57 kahrl in the OnEvent called from guiFormSpecMenu.cpp:809 22:57 kahrl I suppose one should initialize all the members of evt 22:58 kahrl same at guiFormSpecMenu.cpp:883 23:00 kahrl I'll copy the code from guiFormSpecMenu.cpp:971, can I push that? 23:20 kahrl https://gist.github.com/kahrl/6169711 <-- seems trivial enough, builds, fixes the valgrind error 23:20 kahrl pushing in 15 minutes if noone objects 23:21 PilzAdam_ are there any visible bugs resulting from this? 23:22 kahrl no 23:25 kahrl the point of the code is to move the cursor in each field to the end when creating a formspec, and it still does that 23:26 kahrl (I only checked with simple fields, not password fields, as password fields in the default mainmenu always start out empty) 23:29 kahrl (in fact, selecting a world in the "Server" tab resets the password and other fields) 23:48 kahrl WTF, how did this even pass the tests? 23:49 kahrl in filesys.cpp it accesses prefix[prefixsize+len], and prefixsize == prefix.size() 23:49 kahrl filesys.cpp:598* 23:49 kahrl I meant to write prefix[prefixpos+len] 23:50 kahrl changing it doesn't break the tests