Time Nick Message 08:32 MTDiscord I'm not a fan of "send automatically". I know of some applications sending sensitive data in their bug reports. Minetest shouldn't join them. Currently there is a single global log which contains all kind of information (often including chat logs). It should be up to the user to include only necessary log information. 10:02 celeron55 there's also the problem of who hosts the data, and who gets to see the data 14:29 rubenwardy I use Sentry for automatic error reporting, but I added it early on so I know that nothing personal is logged 14:29 rubenwardy Sentry contains tools for censoring info, so you could use that to remove IP addresses and chat messages 14:30 rubenwardy A lot of crashes will probably be things like segfaults, you can't be sure whether any particular memory contains personal info in that case 17:42 erlehmann sentry is really useful, but the vast majority of bugs i saw in sentry was in code that was not covered by tests 17:42 erlehmann besides data collection is a legal nightmare 18:07 celeron55 you have to state next to the "send bug report" button what data is sent, where it is stored, how long it is stored and who is going to see it 18:07 celeron55 i don't think any country disallows it when it's done like that 18:08 celeron55 of course if you then do mistakes and something happens that was not said to happen, then it's an issue 18:08 celeron55 (i.e. data leak) 18:08 celeron55 and you get in the news and are laughed at 18:09 celeron55 a good compromise could be a text field with the report, and a text stating "if you need help, copy this and show it to a project maintainer" 18:11 celeron55 then what happens with the data is up to the user, and the user can choose what messaging systems they trust, and can search and censor the data by themselves as they like 18:11 celeron55 it kind of already happens with our Lua error dialog 18:15 celeron55 it shows the error message, line number, the file with path, and the path generally contains the game and mod name and often even the minetest version, and the traceback 18:15 celeron55 in a copyable text box 18:17 celeron55 or screenshottable 18:17 celeron55 to improve it, maybe a log excerpt could be added, and maybe the engine could attempt to show it even for C++ crashes (it's often possible, altough sometimes it only causes an even more confusing crash) 18:46 Krock sfan5: regarding #11678: do you know whether only large packets are affected? the report repeats seqnums in range [65503, 65531] which indicates a large packet 18:46 ShadowBot https://github.com/minetest/minetest/issues/11678 -- Can not play(or host) game on Android 8&10 for latest master. 18:46 sfan5 it affects everything 18:46 sfan5 also uh seqnums do not correlate with packet size 18:46 Krock there was also a little behaviour change regarding the timeout-reset which was newly introduced 18:47 Krock for reliable packets, there's usually multiple sequntial seqnums 18:47 Krock hence I assumed those would belong together (all channel 1 as well) 18:47 sfan5 ah, could be 18:49 Krock or there might be timeout on client-side for reliable packets that didn't all arrive in time. previously, timed out packets would be re-sent basically instantly 18:50 Krock whereas the time is now reset after every timeout 18:53 sfan5 I don't know what you are referring to 18:55 Krock nvm. invalid idea.