Time Nick Message 00:05 est31 man, we'd need a version of this for minetest servers: https://github.com/thoppe/orthographic-pedant 00:05 est31 "Its diamond, not doamind!" 00:32 est31 okay old pr #2604 00:32 ShadowBot https://github.com/minetest/minetest/issues/2604 -- Add support for multiple listen addesses by ShadowNinja 00:32 est31 it has been rebased in august and it suprisingly still applies 00:33 est31 what to do with it 00:34 eugd add it 00:34 est31 hmmmm, i cant remember, was it approved by you? 00:34 est31 back then 00:36 eugd hah 00:36 eugd oh that always confuses me 00:36 eugd bad username 01:03 ShadowNinja est31: I like eugd's idea :-P 01:04 ShadowNinja nrz and kwolekr have sorta approved it. 01:05 ShadowNinja The CLI argument could probably be easier to use (eg, "--listen foo:bar --listen baz:qux"), but Settings doesn't support that syntax. 01:16 est31 still thinking that one tab is better https://github.com/minetest/minetest/pull/2604#discussion_r27774793 01:16 est31 if style change, then the new style should be correct 01:23 est31 can you fix that? 01:23 est31 and if you have time, what do you think about #3230 01:23 ShadowBot https://github.com/minetest/minetest/issues/3230 -- Better gettext support for protocol version mismatch messages by est31 02:21 ShadowNinja est31: I don't see why it should be changed. 02:21 est31 1. translators can't cope with it 02:21 est31 2. our translation system can't cope with it (xgettext) 02:22 est31 xgettext cann't recognize fgettext(param and "a" or "b") 02:22 est31 can't* 02:30 est31 right now it shows up as "Protocol version mismatch, server " 02:30 est31 so this is more than just a style thing or so, but an actual bug 02:40 paramat EUGD please can you reduce the amount of join/leave message spam in this channel? 02:42 paramat it might be an auto-connect thing :) 03:05 EUGD it totally is an autoconnect thing 03:05 EUGD i'll just leave though 03:06 ShadowNinja est31: I ment the one tab thing. 03:07 est31 ShadowNinja, isnt it code style to only use one tab for indenting long function calls? 03:10 est31 ShadowNinja, usually I don't care, but this is a PR filled with minuscle changes like this: 03:10 est31 https://github.com/minetest/minetest/pull/2604/files#diff-82774cb1985a3f757730324fadbbb49fR27 03:10 est31 or replacing if (variable == false) with if (!variable) 03:15 est31 and I dont think it really looks better with two tabs 03:15 est31 why have you changed it the first place? 03:16 est31 I mean, you removed tabs, but why to something that still isn't allowed by code style? 03:17 est31 but most changes make situation better 03:36 OldCoder est31, thank you for earlier remarks. There is in fact a bug but I will document it before pressing the point. The bug is that auth.txt entries are written inconsistently. Equivalent but different hashes are computed at different stages. This causes authorization to break in some contexts. 03:37 OldCoder As an example of context, if one uses /setpassword, this seems to write auth.txt in a canonical way. However, joining a server for the 1st time may write auth.txt in the canonical way or in other ways. 03:37 est31 yes, that's expected 03:38 OldCoder Go on; interested and thank you 03:38 est31 /setpassword is meant to write old style hashes 03:38 OldCoder If this is expected, what is the appropriate way to check authorization? 03:38 OldCoder If the hashes can be either old style or new style, I mean 03:38 est31 you mean sharing the password you use on the server with other places, like a forum, or an irc bot? 03:39 OldCoder IRC bot 03:39 OldCoder It seems a core issue BTW regardless of mod application 03:39 est31 well, yes, that can be seen as issue 03:39 OldCoder Happy to go play with it... 03:39 OldCoder But is there a way to confirm authorization if hashes may be either style? For example... 03:39 OldCoder is there a get_newhash and a get_oldhash so one can check both? 03:40 est31 but the new style hashes employ a complex protocol in order to protect the user's password 03:40 OldCoder Indeed; see question above 03:40 OldCoder Also; if this is the case, why are 90% of the hashes still created as old style? 03:40 est31 now if you send that password to the server, in the clear, the original sense of the new protocol is lost 03:40 OldCoder Genuinely interested 03:40 est31 they are created as old style because of old clients 03:40 OldCoder Ah! 03:40 OldCoder This is partly client side? 03:41 est31 yes 03:41 OldCoder Got it... 03:41 est31 all clients before 0.4.12 only create old style hashes 03:41 OldCoder So, is authentication by mods still possible? 03:41 est31 0.4.13 clients create new hashes 03:41 OldCoder Got it 03:41 OldCoder Sounds good... but 03:41 OldCoder Is authentication by mods still possible? 03:41 est31 no 03:41 est31 not if using the same password 03:42 OldCoder Very well. So a server that wishes to use mods which authenticate must disable new style hashes? 03:42 OldCoder Is there a line to patch or a setting for this in the core? 03:42 est31 yea 03:42 OldCoder Is it all right to inquire the line or setting? 03:42 est31 install 0.4.12 03:42 OldCoder Um... yes 03:42 OldCoder That is what I run 03:43 est31 on the server side ofc 03:43 OldCoder For server and client 03:43 est31 client can still be 0.4.13 03:43 * OldCoder uses git-curren 03:43 OldCoder t 03:43 OldCoder If I am running git-current 0.4.12 on the server side 03:43 OldCoder already; is this issue consistent? 03:44 est31 you can also replace the line in src/network/networprotocol.h with #define LATEST_PROTOCOL_VERSION 24 03:44 est31 but it may lead to weird bugs 03:45 OldCoder It is part of the protocol at that level? 03:45 OldCoder Hm 03:45 est31 better and cleaner is to downgrade outright to 0.4.12 03:45 * OldCoder submits that this feature, the ability to authenticate, is crucial for some mods 03:45 est31 or to write separate authentication 03:45 OldCoder Does 0.4.13 support old-style hashes? 03:45 OldCoder I.e. via setpassword ? 03:45 est31 you can e.g. do /set_irc_password 03:46 est31 then store the password somewhere in a txt 03:46 est31 then compare that 03:46 OldCoder You propose that mods that need to authenticate write their own authentication, then 03:46 OldCoder Does 0.4.13 support /setpassword with old hashes? 03:46 est31 yes, but that support will possibly be dropped 03:46 OldCoder Hm 03:47 OldCoder One consistency question that is reasonable; I asked it above... 03:47 OldCoder I am running git current 03:47 est31 what do you mean with "consistent" 03:47 OldCoder Isn't that still 0.4.12 ? If so, why am I observing the issue? 03:47 OldCoder Or is git-current 0.4.13 now? 03:47 est31 git current is past 0.4.13 03:48 OldCoder Past? 03:48 * OldCoder thought the screen said 0.4.12 03:48 OldCoder So I'm running at least 0.4.13 already? 03:48 OldCoder This about covers it, I think 03:48 OldCoder Hm; so, no shared authentication for mods 03:48 est31 does it say 0.4.12 or 0.4.12-dev 03:48 OldCoder Moment 03:49 * OldCoder joins one of his worlds 03:49 OldCoder My client says 0.4.12 but the server says 0.4.13-dev. I believe this covers everything. 03:50 OldCoder Well, at least I know that most of my mods work with 0.4.13 now 03:50 OldCoder Hm. Separate question and that is all for now. I have a network patch by Sapier and me... 03:50 OldCoder that is nearly a year old and fixes major problems. It still applies to 0.4.13. So, the network code is not rewritten. 03:50 OldCoder Would core devs be interested in the patch? 03:51 est31 of course, tell more about the patch 03:52 OldCoder Certainly. Around the end of last year I started to encounter the dread RESEND RELIABLE issue. Megabytes of those messages. 03:52 OldCoder Several others and I composed patches. Sapier reviewed and proposed one of his one. I've been testing a hybrid patch ever since then. 03:52 OldCoder And have not seen the issue return. It is very short. Moment. 03:53 OldCoder It's only about 4 lines added or changed. Perhaps I will do a Pull Request or ask VanessaE or sfan5 to do so. 03:53 OldCoder Would this be appropriate? 03:53 OldCoder 03:54 OldCoder one of his own * <- typo fix 03:54 est31 well, sometimes the best fixes are single line 03:54 est31 best you show it first though 03:55 OldCoder Indeed. I will post the patch now, then. Moment. 03:55 OldCoder http://minetest.org/netfix.txt 03:55 OldCoder ^ patch tested for a year 03:55 OldCoder 03:56 * OldCoder has a 2nd optional patch to be shown another time 03:57 est31 yea it would be a good idea to have it as pr 03:57 OldCoder All right. Will look into it. This covers things for tonight. Thank you for your time. 03:57 OldCoder 03:57 est31 but I dont really understand the code of connection.cpp right now 03:57 OldCoder Does anybody? :P 03:58 OldCoder With this patch, mobile is less likely to take down servers. Which is what was happening when a significant number of users connected. 03:58 OldCoder 04:49 paramat will merge #3234 soon in case anyone wants to look ( hmmmm ) 04:49 ShadowBot https://github.com/minetest/minetest/issues/3234 -- Fractal mapgen: Add seabed and large pseudorandom caves by paramat 04:52 hmmmm what's the point of all those casts you added? 04:53 hmmmm when the majority of a file is yours, you just put your name in the copyright 04:53 ShadowNinja !tell est31 The indent for function call continuations doesn't seem to be specified, it just says "placed substantially to the right" which would support two tabs. 04:53 ShadowBot ShadowNinja: O.K. 04:53 hmmmm if two people substantially worked on it, then you just copy the copyright line and use your name 04:53 hmmmm like hud.cpp is an example of this 04:54 hmmmm ShadowNinja, are you ever going to rebase the logging PR?? 04:54 hmmmm I really want to add some new logging features but I can't write over your changes 04:55 paramat ok. casts are an attempt to fix the weird bug 04:55 hmmmm they do nothing 04:55 hmmmm in fact they might mislead people into beliving the variables you casted to a float aren't floats already 04:59 ShadowNinja hmmmm: If it's ever going to get merged. Do you approve of it in it's current state? 04:59 hmmmm I think the last time I looked at it everything was fixed 04:59 hmmmm I'd need to do a once-more over 04:59 hmmmm maybe it'd be a good idea to break it up into separate commits 05:00 * ShadowNinja hopes the conflicts are minor. 05:00 paramat hm yeah they shouldn't be needed apart from casting x y z as floats, further testing will probably show they don't help 05:00 hmmmm this is really hard having things in a large wad like that 05:02 paramat ShadowNinja sometime can you add comments to game#698 ? 05:02 ShadowBot https://github.com/minetest/minetest_game/issues/698 -- New header image 'Minetest Game' by paramat 05:04 paramat i'll remove those unnecessary floats, edit the credits, then merge 05:04 ShadowNinja paramat: Done, LGTM. 05:05 paramat thanks 05:05 ShadowNinja paramat: Also, something similar for textures/base would be good. 05:06 paramat ok 05:11 ShadowNinja hmmmm: I'm not sure what I'd split out. I can add changes in new commits though. 05:12 ShadowNinja I'll rebase it tomorrow (if I remember and have the time). 05:36 est31 ShadowNinja, can't find something that documents the one tab rule, but we really should document it. 05:36 est31 either way, if you really want it that way, you can keep it 05:37 paramat http://i.imgur.com/go8Bz5K.png fractal and sandbank ( hmmmm ) 05:41 paramat now merging 3234 05:48 paramat done 15:53 rubenwardy do people read this: https://bugs.launchpad.net/ubuntu/+source/minetest ? 15:53 nrzkt no 15:57 rubenwardy just pointing it out 23:20 ShadowNinja hmmmm: Logging PR rebased.