Time Nick Message 15:22 rubenwardy So my understanding is that Client handles the networking and state of the client, and Game is for the rendering and user input. The relationship is Game -> Client. But why does client->makeScreenshot() exist? Surely this is a Game responsibility 15:23 rubenwardy hm also has references to client and rendering engine 15:23 rubenwardy I suppose the latter is needed for loading textures from the network 15:23 sfan5 isn't the Game class hidden from outside so that Client may need to have such wrappers? 15:24 rubenwardy hm 15:24 sfan5 but I doubt the distinction is so clear cut anyway 15:24 rubenwardy makeScreenshot is called from Game and GUIFormspecMenu 15:24 rubenwardy so I guess GUIFormspecMenu couldn't call Game, but we do have screenshot.h 15:46 [MatrxMT]_ Am I the only one with broken GitHub? It's been at least an hour, PRs not loading and such 15:46 MTDiscord not just you, big reds everywhere: https://www.githubstatus.com/ 15:47 [MatrxMT]_ Lovely 15:49 red-001 yes I mentioned this on discord but the uptime for Github has been terrible lately 15:49 red-001 perhaps discussing a different host should be reopened at this rate 16:11 MTDiscord reopened? lol 16:11 MTDiscord at least iirc there is an issue for that still open 16:17 [MatrxMT]_ yep #12324 16:17 ShadowBot https://github.com/luanti-org/luanti/issues/12324 -- Eventually move to a free git platform 16:21 red-001 oh I see it was reopened already from #7412 16:21 ShadowBot https://github.com/luanti-org/luanti/issues/7412 -- Move repository 16:22 MTDiscord heh, im the second to last comment there. in a year, think ive soured a bit on codeberg. self hosting would probably be ideal, but i dont know if its realistic. git forges seem to be attacked to death a lot these days, and even big distros disable signups, etc. also luantis history of hosting things all in is rather poor 16:23 red-001 ah yes the forum 16:23 sfan5 self-hosting is annoying even with 100%. let's at least choose a platform that has some established user base 16:23 sfan5 also so we can continue to get free CI 16:24 red-001 gitlab 16:24 MTDiscord forum is crap, cdb has slow issues, serverlist probably is the best. solution: have sfan5 host everything 😛 16:24 sfan5 s/100%/100% uptime/ 16:24 red-001 I feel CDB should not just send one massive mod list on connect maybe 16:24 MTDiscord gitlabs design is ass tho. also the project already used it previously and ditched it due to ci issues 16:25 red-001 when was the gitlab era 16:25 MTDiscord like 5 years ago? it was used for ci, docker builds 16:25 sfan5 "used" as in we had an active mirror and used their CI 16:25 sfan5 yeah 16:25 red-001 yeah Nrz set that up I think but it wasn't like used-used 16:26 red-001 what was the issue with their CI? 16:26 MTDiscord they wanted cc details? i think? 16:26 MTDiscord that was the final nail in the coffin 16:27 red-001 oh I don't think I ever had that issue 16:27 sfan5 feel free to search the logs but I remember the issue being that they restricted their free CI hard 16:27 sfan5 I believe that was during the time when mining monero on CI became popular 16:27 MTDiscord certain someone ranted about cc vs debit cards in europe 😛 16:28 MTDiscord im my recollection is correct 16:28 red-001 lately I have been soft banned from google search for some reason so I imagine I won't get any search results if the irc logs still are searched via google link 16:28 sfan5 you can use any search engine to search the logs of course 16:28 red-001 it's very annoying if I use any search specifier google forces me to go through like 2 bot verification pages, and duckduckgo just doesn't have the same quality of search results 16:29 MTDiscord are you not signed into a google account? 16:29 red-001 it does keep signing me out of that as well 16:29 MTDiscord ive found more and more stuff starts forcing more verification if you arent signed in, even if its free to use not signed in 16:29 Krock perhaps time to clear some cookies, cache and hope that the issue is resolved afterwards 16:30 MTDiscord which, given the time we live in, makes sense i guess 16:30 red-001 probably cause of AI, the verification pages that is 16:31 user333_ I have a local copy of the IRC logs that I grep as needed 16:31 user333_ or at least, had 16:31 red-001 logged in 16:31 red-001 still got bot blocked 16:32 red-001 but the fun thing is I found out google has 3 different bot block pages for search 16:32 red-001 at least 16:32 red-001 not really sure why 16:34 red-001 Maybe gitlab got better now? 16:34 red-001 or they would generously donate some CI time 16:35 red-001 I have no particular love for them but Github has horrible up time lately 16:38 MTDiscord i think most git solutions are better than gitlab. reason all the forges copy github more than gitlab in terms of design 16:39 [MatrxMT]_ I still think Codeberg is the best solution out there, plus their community is very kind and their staff can help us if needed. I don't know about CI warts and all though 16:39 user333_ my preferences would be (a) a selfhosted forgejo, which has a seperate selfhosted CI to go with it. not to mention i hear they're working on federation so people from other instances (that will include codeberg) will be able to contribute (b) move to a large existing forgejo instance like codeberg. though i am not an active developer at all so my 16:39 user333_ opinion probably doesn't matter too much 16:40 MTDiscord codeberg has already altered the deal once, i wouldnt be surprised if they do it again (ai policy) 16:40 MTDiscord I don't think codeberg's AI policy is much of an issue IMO 16:40 MTDiscord the federation idea, while cool, as far as i am aware (disclaimer), hasnt really gone anywhere significant in the past # years since it started 16:42 MTDiscord for luanti? probably not. more so just pointing out github/gitlab are kinda do wherever (within reason), whereas codeberg has said no, not neutral 16:43 user333_ yeah that is a major bottleneck for anyone who wants to contribute, they would have to have an account on each instance vs one account for every instance with federation 16:47 rubenwardy Codeberg's AI policy is a non issue, its way less strict than ours 16:47 MTDiscord yeah codeberg also previously restricts repositories for personal/private/non-community projects with certain limitations on their size. it's an opinionated platform which seeks to host a certain slice of projects that align with the goal of their non-profit organisation, but luanti is well within that range 16:47 rubenwardy We have good connections with the codeberg staff and could become members of the eV 16:49 MTDiscord there is a roadmap, still seems a long way off https://codeberg.org/forgejo-contrib/federation/src/branch/main/FederationRoadmap.md 16:49 red-001 what's their CI situation? 16:49 sfan5 i will state my opinion again for emphasis: please no self hosting, and don't fall for "federation will fix everything" either 16:51 MTDiscord you need to request access i think? they have Woodpecker CI and alpha forgojo actions 16:51 MTDiscord the latter is more the "Standard" yaml that github and gitlab have 16:52 MTDiscord https://docs.codeberg.org/ci/ 16:52 red-001 I had 99 problems, so I joined a federation, now I have a 100 problems all of them distributed across a network of loosely cooperating third parties with potential eventual consistency 16:53 MTDiscord once nice thing about codeberg is they host weblate, so could be an all in one solution 16:54 MTDiscord another point about codeberg's own provided CI runners is that they are AMD64 linux only. no native ARM runners, no macOS and no windows runners 16:54 MTDiscord last i checked however, there pages/hosting was not recommended. however i have offered to host the docs website 16:54 red-001 I'm not sure it really sounds usable, I see a lot of warnings on that page, giving up Windows MSVC builds would be sad but in general it seems very limited 16:55 MTDiscord a way of framing it might be, is it realistic to be on codeberg if github was a mirror? keeping github as a mirror and for ci 16:56 MTDiscord yeah, that would be possible. you could just push commits from codeberg back to github and github actions would run on those commits 16:56 red-001 would that work for PRs? 16:56 MTDiscord not ideal, but..... the project has done it previously 16:57 sfan5 i think we can live with no msvc, no arm and no macos 16:57 MTDiscord it would not work with PRs. so basically only anything that ends up in master 16:57 red-001 gitlab was more of a nice to have CI workflows, not the core ones 16:57 sfan5 if any of those are considered essential we could use project funds to buy third-party CI (that can integrate with codeberg) 16:57 red-001 I suppose it just depends on how acceptable breaking MacOS/Arm in a PR by accident is 16:58 red-001 well aren't unit test exit codes ignored by CI right now 16:58 sfan5 no 16:59 MTDiscord it does look like codeberg supports being a container registry. however it might be a good idea to recommend a luanti domain name in front of it or github, so if the project ever moves again 16:59 sfan5 only in the msvc CI run because nobody has bothered to fix it 16:59 sfan5 not to mention that the CI run itself is constantly broken 16:59 red-001 freedesktop needs a new load balancer or something 17:07 sfan5 soooooooo 17:07 sfan5 do we have any kind of consensus on releasing on 19th aug? ping: rubenwardy @luatic Krock 17:08 sfan5 there's only #17348 (fix, could be moved to next release) and #16896 (needs action) left in the milestone 17:08 ShadowBot https://github.com/luanti-org/luanti/issues/17348 -- GUIChatConsole: Fix (smooth) scrolling by arrows by SmallJoker 17:08 ShadowBot https://github.com/luanti-org/luanti/issues/16896 -- Severe rendering regression when array textures are used 17:08 Krock sfan5: need to review some of your PRs until then but otherwise - seems doable. 17:08 sfan5 oh yeah of course 17:08 sfan5 I'll put them in the milestone 17:09 Krock plus the other settings-related patch that yet needs a PR for 5.17.0 17:09 sfan5 yes 17:14 sfan5 Krock: these commits are essential IMO https://x0.at/YydM.png 17:22 Krock Green Is My Pepper 17:30 Krock sfan5: does generate_from_settingtypes.lua work for you? I am using RUN_IN_PLACE=1 but out of tree. minetest.conf.example can no longer be written. 17:31 red-001 umh hmm 17:31 red-001 that's not good 17:31 sfan5 https://github.com/luanti-org/luanti/issues/16079 17:31 Krock perhaps it's differently in-free when the execution directory is not bin/ ? 17:31 Krock ah 17:31 red-001 sfan5 btw sorry about the git show, it's been a while since I had to make a patch file completely forget that whole git show vs git format-patch distinction, idk if that's on the patch list 17:32 red-001 I uploaded a new patch with that resolved, it should cleanly apply 17:32 red-001 Krock is it trying to write to in-tree? 17:33 sfan5 it might actually even still work with an in-tree build 17:34 Krock in-tree: Blocked attempted write to minetest.conf.example" 17:35 Krock though does work when the execution dir is the project root 17:35 red-001 sandboxing is here to protect you from yourself 17:35 sfan5 I assume you're re-generating it to get rid of the "fake" changes I included? please commit that to master 17:51 Krock GOSHDARN IT GITHUB 17:52 red-001 oh 17:52 red-001 did it go down again? 17:52 Krock discarding my review notes just by changing to the next commit 17:52 red-001 amazing UI 18:01 red-001 Krock when you have a second, could you see the sample code and update I posted on the github gets routed to the right place, I'm not sure I have visibility on the correct discussion but I think it's a somewhat relevant update if it hasn't been independently noted. 18:13 Krock sfan5: is it any issue if builtin code can be `dofile`'d from a mod during load time? 18:14 sfan5 no 18:14 Krock the worst case that could happen is leaking an internal state, I guess. 18:16 sfan5 i'm not sure if builtin even even still has special permissions inside the sandbox, but these cease to exist before the first loadMod() is called 18:46 Krock red-001: which sample code exactly? the one with get_entity_name ? 18:47 Krock I don't see any open Security Advistory for this specific UAF issue 18:49 red-001 okay Lars just said something about it being the wrong issue so I figured maybe there was one, I just did the write up of the user spacing version in that existing GHSA in case yous want to put it out, can discuss the details on Github if needed considering the release is not quite out yet 18:55 red-001 if there isn't a generic GHSA I don't know which one I was meant to post that information on 18:57 red-001 GHSA-24fm-82f9-hc6m has a fix patch for review attached btw even if I don't really know if it needs it's own advisory in the end 19:35 sfan5 rubenwardy: any news about requesting the sdk update deadline extension for android? 19:37 rubenwardy https://rwdy.uk/WGfc1.png 19:37 rubenwardy #2 ? 19:37 ShadowBot https://github.com/luanti-org/luanti/issues/2 -- Burned wood 19:38 rubenwardy or 1 19:39 sfan5 I would pick 2 19:39 rubenwardy done 19:39 rubenwardy seems to be autogranted 19:41 sfan5 neat 21:35 MTDiscord sfan5: don't we want #17448 in 5.17.0 too? 21:35 ShadowBot https://github.com/luanti-org/luanti/issues/17448 -- [no squash] Use-after-free fixes by sfan5 21:36 sfan5 yes 22:58 sfan5 I'm planning to merge the release-essential commits (as marked above) from #17442 tomorrow 22:58 ShadowBot https://github.com/luanti-org/luanti/issues/17442 -- [no squash] Misc fixes and filesystem stuff by sfan5 22:58 sfan5 (tomorrow as in 2026-08-18)