Time Nick Message 00:21 luk3yx Can I make a confidential bug report or something? 00:25 luk3yx I found a potential issue but I don't want a whole bunch of swines trying to exploit it 00:27 paramat yes, are on the MT forum? if so send me a private message and i'll make sure it gets seen by other devs 00:29 luk3yx Okay 00:33 luk3yx paramat: PMed 01:00 paramat thanks 01:03 paramat got it 01:05 luk3yx paramat: Will it be fixed in 5.0.0? 01:26 paramat heh nuuu 01:26 paramat too late, release may be within 2 days 01:27 paramat it's not a critical enough bug to delay release 01:31 IhrFussel Why do people pay hosters MANUALLY in 2019? This is what happens quick then https://forum.minetest.net/viewtopic.php?f=10&t=22033 01:31 rubenwardy OVH doesn't offer automatic payment 01:32 GreenDimond Yes it does 01:32 rubenwardy https://i.rubenwardy.com/XaLRr.png 01:33 rubenwardy oh, well it's just kimsufi 01:33 rubenwardy OVH comes up on my bill, the payment page says OVH, yet the renewal is different 01:35 GreenDimond where is pasteboard when I need it 01:36 GreenDimond https://pasteboard.co/I3u8F7W.png 01:36 rubenwardy oh, well it's just kimsufi 01:37 GreenDimond ? 01:37 rubenwardy Kimsufi is a subsidary of OVH 01:37 GreenDimond Ah 01:37 rubenwardy it's OVH, but branded differently 01:37 rubenwardy OVH comes up on my bill, the payment page says OVH, yet the renewal is different 01:38 GreenDimond Do you use OVH or Kimsufi? 01:43 VanessaE the solution is simple: at some point, pay two months' worth of service in one month, then continue paying the normal monthly charge. 01:44 VanessaE that way you have a big time buffer 01:44 VanessaE that's what I did to avoid that same problem 01:45 VanessaE and with OVH servers, you can transfer the remaining time from one server to another including between e.g. SoYouStart and Kimsufi 01:45 VanessaE (which is what I did when I migrated over the past couple days_ 01:45 VanessaE ) 01:46 VanessaE so my next bill isn't due until like, April 28 or so. 01:46 rubenwardy Oh really 01:46 VanessaE yeah 01:47 VanessaE but to avoid using up that buffer, I'll just pay it on or around April 1 01:47 VanessaE I do pretty much the same thing with my phone bill 01:47 VanessaE (it's a prepaid, so it has very much the same property) 01:50 GreenDimond Oh, you use OVH as well? 01:50 VanessaE yes. 01:51 GreenDimond I feel like I chose the right host 01:51 VanessaE they're a good host. 01:52 GreenDimond pretty flipping cheap too 01:52 VanessaE mmhmm 01:52 GreenDimond I paid $48 for a year of hosting on a fairly minimal machine (4GB RAM, 1 core, 40GB Disk) on a black friday deal 01:53 VanessaE someone clued me in to Kimsufi's prices (didn't know they'd gone down) relative to SYS, so I went for it. 01:54 GreenDimond whats the diff between Kimsufi and OVH? 01:54 VanessaE just different markets, really 01:54 VanessaE KS servers are a bit lower-end 01:54 GreenDimond looks the exact same to me x) 06:26 luk3yx Is it safe to post that issue onto GitHub (as it's "non-critical") along with some lua code to work around the issue?? 06:26 luk3yx s/\?\?/\?/ 06:26 luk3yx Bad sed expression 06:32 luk3yx .help tell 06:32 luk3yx ~help tell 06:32 luk3yx !tell 06:32 MinetestBot luk3yx: Need a nickname... 06:33 luk3yx !tell paramat Is it safe to create a GitHub issue for that bug/exploit (along with lua code to work around it)? 06:33 MinetestBot luk3yx: I'll pass that on when paramat is around 06:35 sofar in general, if you find a possible exploit, you should describe it, but not post the literal exploit if it is complex 06:35 sofar if you have a fix, post that for sure 06:36 luk3yx It's a hacky fix 06:36 sofar luk3yx: also for sure PM any of the devs and they themselves can share it 06:36 luk3yx I did 06:36 sofar make a secret gist and i'll post it to the other devs as well 06:36 luk3yx http://irc.minetest.net/minetest-hub/2019-03-02#i_5503146 06:37 sofar is it mtg or mt? 06:38 sofar well, up to you, but I'd appreciate if you would PM me as well 06:38 luk3yx It's Minetest itself, but sure 06:38 luk3yx Let me compile a gist with a description and fixes 06:53 luk3yx sofar: PMed you (in case you didn't get it) 07:58 Krock https://github.com/minetest-mods/unified_inventory/pull/123 07:59 Krock ^ something interesting for IhrFussel 08:13 Krock now with speed test results. this was totally worth it. ~10 times faster execution 08:14 Krock well so, I reduced the stack movement function execution time, as well as the loop complexity. yay 08:21 * Krock triggers VanessaE who's sleeping right now 10:06 nerzhul )nice job Krock 10:09 Krock thank you, nerzhul :D 10:41 Krock o/ kaeza 10:47 * VanessaE looks at Krock 10:47 VanessaE you twit, you woke me up. :P 10:48 VanessaE but, I don't have write access to the UI repo. 10:49 Krock well, I do. but you might be interested in testing it :P 10:49 Krock I hope you slept well 10:49 Krock (until nasty me came) 10:49 VanessaE heh, yeah slept okay. 10:49 VanessaE kinda up early though.. 10:49 Krock !mod mail 10:49 MinetestBot Krock: In-Game "Email" [mail] by cheapie - https://forum.minetest.net/viewtopic.php?t=14464 10:52 VanessaE anyway, better I should leave testing to others. 10:52 VanessaE my server box is still "In a state of flux" as we say here. 10:53 VanessaE dust hasn't settled yet, won't be until next thursday 10:53 VanessaE (but I'll still pull in updates as needed, so if someone merges that, I'll get it then) 11:19 Krock oh, you had server issues? 11:19 VanessaE nah 11:20 VanessaE someone pointed me at a lower-priced host (Kimsufi, another OVH subsidiary), so I chose a box, using my server stats page as a guide to how much oomph I need. it's a cost-saving measure - the new box is a little over 1/3 the price of the old one. the upside, other than price, is this one has way more storage, and it's in canada instead of france. 11:21 VanessaE and G*d knows, I need all the money I can get :P 11:21 Krock canada instead of france is a downside, at least for me :P 11:21 Krock good to hear that you found this optimization 11:22 VanessaE yeah and speaking of which, I have to wait until thursday for a full week's worth of backups to accumulate, before rsync reaches its best performance 11:23 VanessaE only thing left then will be disk-IO-bound (minetestmapper) 11:24 VanessaE (I guess the mapper chews through the entire map database instead of seeking/pulling db records for just the parts of the world that need drawn) 12:26 nerzhul VanessaE: scaleway is french too and cost less :) 12:26 nerzhul but OVH is a great competitor 12:26 nerzhul you can have 3 euros per month 2 gb 4 cpu servers with ARM technology and they work great 12:26 nerzhul i have 10 servers like this :p 12:34 VanessaE ARM ain't enough for a heavy MT server that runs technic/mesecons/pipeworks/diglines :P 12:35 VanessaE nor would 2GB be enough for my workload anyway 12:36 rubenwardy arm does a whole range of stuff 12:36 rubenwardy it dominates the power efficiency market 12:36 rubenwardy ie: phones and embedded 12:36 VanessaE but the one I have now is $17/mo. 2core/4thread i3-2130 3.4 GHz, 8GB RAM, 2TB HDD, 100 Mbps net 12:37 rubenwardy mine is 12:37 rubenwardy Intel i3-2130 2c/4t 3.4GHz 8GB DDR3 1333 MHz 1TB 100 Mbps /128 €14.99 ex. VAT 12:37 VanessaE rubenwardy: yeah I know.. my printer runs on a 100 MHz Cortex M3. 12:37 rubenwardy oh lol, it's the same one 12:37 VanessaE heh 12:37 VanessaE yep, the KS-7 model 12:38 rubenwardy noise 12:38 rubenwardy well, I got the 1TB rather than the 2TB 12:38 rubenwardy was slightly cheaper at the time 12:38 Calinou ARM on servers is still niche right now 12:38 rubenwardy Linux has some things to say about that 12:38 rubenwardy mostly stupid things 12:38 rubenwardy but some things 12:38 rubenwardy *Linus 12:39 VanessaE if I didn't run technic et. al, I'd probably be just fine with the KS-2 package 12:39 Calinou I'd argue kernel developers don't like ARM because of its tendency to release lots of binary blobs :) 12:39 Calinou (ARM GPUs especially) 12:39 VanessaE ARM makes GPUs? 12:39 rubenwardy *arm 12:39 VanessaE TIL. 12:39 rubenwardy yes, Cortex 12:40 rubenwardy fun fact: the internal code names for arm's GPUs come from Greek mythology 12:41 rubenwardy another fun fact: the internal code names for arm v8.5a extensions come from Scottish whiskeys 12:42 rubenwardy on completion of a project, they'd have a little party with the whiskey that the project was called 12:43 Calinou by ARM GPUs, I mean the ones you typically find on Android devices 12:44 VanessaE huh. I thought "Cortex" was just some reference to the brain 12:44 rubenwardy I'd imagine they'd be integrated 12:44 Calinou they are integrated, but they're still GPUs :) 12:44 rubenwardy Cortex is the external facing name 12:46 VanessaE now I have to wonder what the "internal" name is :) 12:47 VanessaE (the controller uses an LPC1768 chip, fwiw) 12:54 nerzhul VanessaE you can have 4 or 8 GB too 12:54 nerzhul https://www.scaleway.com/pricing/#anchor_arm 12:59 VanessaE yeqah... no.. to even get enough storage for my workload would be the 139.99€ package 12:59 VanessaE -q 13:00 VanessaE but the lower end of that scale is pretty good pricing, if your workload is small enough. 13:00 VanessaE glad to see prices are finally coming down.. I remember when the server I now have woulda cost me $100 a month :P 13:05 Calinou common hosts are still expensive to my taste 13:05 Calinou my VPS is from a nonprofit French provider, those can offer really good prices 13:05 Calinou and near-100% uptime so far :) but that's more luck than anything 13:45 Sokomine VanessaE: i still wish for dates to be added to map data...created, last changed, amount of blocks digged and placed by players... 13:48 Sokomine Calinou: isn't it mostly a problem with arm gpus and not with the cpus as such? driver support for gups seems poor 13:51 Calinou Sokomine: yeah 13:51 Calinou mobile drivers are beyond horrible 13:51 Calinou if you thought desktop graphics drivers were poor, look at mobile drivers :) 13:52 sfan5 nvidia supports tegra pretty well though 13:53 sfan5 but that's the exception, not the norm 13:53 Sokomine Calinou: i just know that the raspberry gains a lot of its advantages due to the gpu (to which the smaller cpu seems to be attached?) beeing quite well supported 13:53 Calinou yeah, NVIDIA even has ARM devboards supporting plain OpenGL 13:53 Calinou (i.e. not OpenGL ES) 13:54 Sokomine arm servers might be good enough for a lot of purposes. a small mt server ought to run fine on them 13:55 Calinou maybe, the single-thread performance is usually still pretty bad 13:55 Calinou except on Apple CPUs maybe :) but you won't find those in servers 13:55 Calinou there's people hosting public game servers on Raspberry Pi and we always have to tell them to stop it :P 13:55 Sokomine but then - where does mt use multiple threads? except mapgen (and that may be slow on faster devices) 13:55 Calinou because these often become lagfests if more than a few players join 13:56 Sokomine but as long as it's just a few players and nothing fancy running (wouldn't dare to see ve-s hosted on a pi) it ought to work pretty well. imagine creative mode, with a flat map, no fancy mods... 13:57 Sokomine the pi may suffer a lot from its slow interfaces 15:08 IhrFussel Until recently I used num_emerge_threads '3' on my server with sqlite so the claim that 'anything other than 1 is dangerous' is wrong...my map is fine, I had no weird DB crashes at all ... now I'm testing if my server runs better with it at '1' but even the sqlite docs mention that the default mode is 'serialized' and can handle basically infinite threads with no restrictions 15:09 IhrFussel So unless the core devs f'd up I would say it is safe to set it higher than 1 15:10 IhrFussel https://www.sqlite.org/threadsafe.html 15:10 IhrFussel 'Serialized. In serialized mode, SQLite can be safely used by multiple threads with no restriction.' 15:10 IhrFussel 'The default mode is serialized. ' 15:22 IhrFussel Hi Krock and thanks for the PR ... when it is merged is it enough to just replace my register.lua with the new one? 15:22 IhrFussel Cause I heavily modified the other files 15:23 Krock yes 15:23 IhrFussel But it's not tested yet right? 15:24 Krock if you want to test it, here's the file: https://github.com/SmallJoker/unified_inventory/blob/fast_stackmove/register.lua 15:24 Krock IhrFussel: please read the comment. I tested various cases 15:24 IhrFussel If I replace it on my production server now...what's the worst that could happen? Stacks just not moving correctly? 15:24 Krock basically I re-wrote the entire copy-to-craft-list function 15:25 Krock yes. there's a safety mechanism in case it fails 15:25 Krock it would undo the inventory action (taking items) 15:25 Krock https://github.com/SmallJoker/unified_inventory/blob/fast_stackmove/register.lua#L480-L481 15:26 IhrFussel Sounds good :D Will replace it now and ask my players if they noticed any bugs 15:26 Krock the inventory is only modified in these few lines 15:26 Krock I hope you also saw the stats. 10 times faster execution :D 15:27 Krock plus protection against too high submitted copy-to-craft button fields (CSM + edited client) 15:27 IhrFussel Yeah... I might still change the '99' button to '50' cause sometimes that button lagged my server for almost 20 secs... which means it could still lag it now for 2 secs 15:28 IhrFussel And it should be rare that a player needs a full crafted stack of something 15:28 Krock try it with 99. still possible to reduce it afterwards 15:28 IhrFussel I will 15:28 Krock uhm.. I do. crafting oil extract with chainsaw-cut leaves 15:29 Krock two clicks vs 6 times shift-click :D 15:29 Krock in theory I could extend this feature so that it takes items from additional lists, such as bags 15:30 Krock but then you quickly lose the overview 15:44 IhrFussel unified_inventory.trash_enabled gets set nowhere or am I blind? 15:45 Krock init.lua 15:45 Krock grep -r is your friend 15:47 IhrFussel I used that 15:47 IhrFussel WAIT 15:47 IhrFussel Then my version ofu_i doesn't have it 15:47 IhrFussel So it#s good that I removed it from the file (the check) else it would crash 15:47 IhrFussel Or nobody would see the trash slot 15:48 Krock exists for three years now 15:49 Krock two years actually. https://github.com/minetest-mods/unified_inventory/commit/7ab882546 15:52 IhrFussel Not sure from when my version is...I just replaced necessary files for important fixes/patches 15:52 IhrFussel I guess init.lua never had to be touched 15:52 IhrFussel I think I found a bug 15:53 IhrFussel When you try to craft more than you have it just moves the full stacks into the craft slots and doesn't distribute them 15:53 Krock hmm right 15:54 IhrFussel And it looks like it only crafts one at a time...not sure right now if that was the same before 15:54 IhrFussel Like I have the items for 10 but can only take 1 out of the slot 15:55 Krock huh? Use the scroll button to craft 10 15:58 IhrFussel Okay looks like that was before too 15:58 IhrFussel So the only bug is the moving of whole stacks into 1 slot 15:59 Krock yes.. trying to fix that 15:59 IhrFussel But it only happens when it tries to move more than you got 15:59 Krock ofc 15:59 IhrFussel If you have the amount it works perfect 15:59 IhrFussel Will it just be a few more lines in 1 function? 16:00 IhrFussel Meaning I can copy&paste? 16:00 Krock most likely, yes 16:00 Krock will have to sum up all stacks 16:00 Krock then divide as good as possible 16:00 Krock issue is that a recipe can contain two times the same item 16:02 IhrFussel You could also just abort the action 16:02 IhrFussel That would probably make sense too...maybe even more 18:33 rubenwardy my server ip is 192.168.50.118 18:33 rubenwardy !up 127.0.0.1 18:33 MinetestBot 127.0.0.1:30000 seems to be down 18:33 rubenwardy sadness 18:55 Krock IhrFussel: I tried many things but figured out that stepwise movement is the only sane solution 18:55 Krock any other produced more than 200 additional lines 18:55 Krock and wasn't much more efficient 18:56 Krock https://github.com/minetest-mods/unified_inventory/pull/123/commits/fa69670f 18:57 Krock copying steel block recipe this way results in just a millisecond execution time of the code. almost the same as before (~800 microseconds) 18:57 Krock varies a lot depending on the Minetest/CPU load 18:59 Krock I tried to cache the results, but non-integer indexing of tables sent all efforts straight into the next trash bin 19:00 IhrFussel Okay, replaced the function now in my file and will restart in a bit...then report results :P ... thanks for your hard work 19:00 Krock np. I really tried to make this better the last three hours.. these three changed lines are all I can offer 19:01 Krock at least I learned a lot about the Lua and C bindings performance 19:05 Krock wow, stu is providing some top-notch animations https://vimeo.com/319038756 :D 19:06 Krock needs to enable the rotation interpolation, though. 19:08 sofar wrong link? 19:10 Krock https://vimeo.com/320857098 19:11 Krock vimeo apparently also has autoplay. copied the next video in the loop 19:15 tenplus1 hi folks :) 19:15 Krock hi tenplus1 19:15 tenplus1 hey Krock 19:15 tenplus1 snap! :P ehehe 19:15 Krock uuh cloose 19:15 Krock faster this time. by a fraction of a second 19:15 tenplus1 :P 19:16 Krock :P how are you? 19:16 tenplus1 okies thx, updated contentdb page and added a few new mods... you ? 19:17 Krock oh nice. continuously developing :) 19:17 tenplus1 arent we all :D 19:18 tenplus1 some of the releases needed renewed 19:18 Krock I spent some hours in optimizing the copy-to-craftgrid function of u_i. It's a lot faster now, but has a little different behaviour 19:18 tenplus1 ooh nice... 19:18 tenplus1 you tried kilbith's latest craftguide ? 19:18 Krock but now copying 10x or 99x doesn't result in any noticeable speed difference 19:19 Krock no, I'm by default a u_i user, didn't give the other inventories a big chance unless when I played on servers 19:19 Krock is there a cool new feature? 19:19 tenplus1 always felt ui was slow for me... er, he's redone a few functions so it a LOT faster and clicking on item can switch you to other recipes using it 19:20 tenplus1 we got it running on xanadu 19:20 Krock oh I see. already exists in u_i but it's nice to have it in craftguide as well 19:21 tenplus1 :P 19:21 tenplus1 I did add my fork of builtin_item and falling_item to cdb for players to check out 19:22 rubenwardy have you seen your notifications on CDB? 19:22 tenplus1 hi ruben, lemmie check 19:23 tenplus1 ooh the bell is lit up :D 19:23 tenplus1 hi calcul0n 19:24 tenplus1 my stamina fork has the differences in the long description 19:25 tenplus1 tweaked 19:28 tenplus1 er, mobs_npc was written alongside mobs redo, it's the original 19:29 tenplus1 the other mod in question is called aliveai and not mobs_npc, it just appears when you do a search tho 19:40 tenplus1 all done, thanks ruben :D 20:16 tenplus1 cya folks 20:32 Krock IhrFussel: did you already have time to put it into test? 20:32 IhrFussel Nope, will restart now 20:33 Krock > when you want to join a 0.4.x server but your 2 month old CSM causes a crash 20:34 Krock til I haven't played on 0.4.x for a very long time 21:15 kaeza Greetings. 21:15 kaeza o/ Krock 21:16 Krock \o 21:25 IhrFussel sofar, why do you assume that sqlite is not thread-safe again? The docs say it IS by default...so to not be thread-safe the devs must've MANUALLY changed the mode 21:26 IhrFussel And I'm pretty sure my server would've segfaulted a lot cause of mapgen then cause I set my setting to '3' 21:31 rubenwardy it's the default assumption with C++ 21:40 IhrFussel Well first I believed that too when I saw the github issue about the warning...but then someone linked the docs that clearly state default mode is 'serialized' so that usually cannot cause segfaults