03:40:52 rbrunner7: Just leaving this here in case you're interested; I'm not sure how accurate of a retelling this is ;) 03:40:54 https://github.com/monero-project/monero-docs/pull/408 04:39:45 Thanks, will have a look later today 16:31:17 an asmap (for outbound connections only atm) is added in https://github.com/monero-project/monero/pull/11474 , its currently 'bring your own --asmap file'. copied and licensed appropriately from bitcoin core. since feb they have moved to embedding the the .dat in their binary. i thought this would be a problem but we do the exact same thing with 16:31:17 checkpoints.dat so its not much 'new' code to add. at release the .dat and hash file is bumped if an update exists @ https://github.com/bitcoin-core/asmap-data/releases 16:32:52 was wondering if there where any opinions reg embedding the .dat file in our tree? the steps on top of 11474 look like https://github.com/plowsof/monero/compare/asmap...test-asmap (which i again hope is licensed appropriately) 16:33:52 I would be in favour of something like this 16:34:16 I'd be okay with updating the .dat file during release. 16:36:22 Forgive me to be critical, but did somebody already check whether using this ASMAP stuff has actual, tangible benefits compared with the current approach? I don't remember whether @rucknium:monero.social tested something, or only had educated guesses. Peer selection is non-trivial, and in some cases the final result "How many of my peers are probably spy nodes" is hard to predict. 16:37:44 plowsof found new suspected spy nodes that bypass our current approach 16:38:49 https://observer.monerodevs.org/d/peersel 16:39:40 As of now, I am not opposed to this as an opt-in flag. IMHO, much more analysis would have to be done to set it as default. 16:40:05 bitcoin is still opt-in currently, after all these years 16:40:18 To set it as default, you would want to have an analysis that shows, with reasonable confidence, that the cure would not be worse than the disease, as I did for the /24 subnet deduplication. 16:40:45 Well, ok, but Bitcoin is hardly a role model of adopting technological improvements ... 16:40:54 :P this is true 16:40:56 An adversary could just spread across many ASNs, in response to a default rollout of this, potentially making the issue worse. 16:40:57 (regarding not yet default there) 16:41:54 There are many honest nodes concentrated in the ASNs of VPS vendors. An adversary could exploit the ASN avoidance logic to accumulate more outbound connections than the honest nodes. 16:41:59 Yes, in a bad case we do effort, change code, and let people update for nothing more than a placebo. Worst case it makes things worse. 16:43:04 ASNs are harder to analyze than subnets because ASNs do not have a defined size. I used the defined size of subnets in my game theory analysis. 16:44:05 Hmm. Maybe it would be nice to have this in the "bag of tools", so if the adversaries switch tactics we have another tool deployed already that people could use on short notice. 16:44:42 Maybe also has some effect as a determent? 16:45:17 so there is an identifiable fleet of about 1.3k nodes who share the same fingerprint. the majority of which are in DigitalOcean 16:45:40 Analysis is here. See section 4 "Protocol-adversary interaction as a game" in particular: https://github.com/Rucknium/misc-research/blob/main/Monero-Peer-Subnet-Deduplication/pdf/monero-peer-subnet-deduplication.pdf 16:46:34 the asmap for outbound connections is not "the solution" it is part of the bag of tools 16:47:22 At least it would ostensibly make the attack much more expensive for them > <@rucknium> An adversary could just spread across many ASNs, in response to a default rollout of this, potentially making the issue worse. 16:47:38 @jpk68:matrix.org: Have you proved that? 16:47:54 No, I'm just spitballing :) 16:49:12 But it seems this entity is currently able to run spy nodes en masse through LionLink, which I can't imagine is too expensive compared to having to use a bunch of different providers 16:49:25 If its is made the default, you would also stress nodes in low-population ASNs by creating many inbound connections to them. 16:49:56 it is opt-in with my proposed patch --asmap embedded or path 16:50:45 Yes. I've said already that I'm ok with it as opt-in, while we wait for better analysis of how the adversary could respond to this. 16:51:27 every defence can be circumvented - the end goa is pushing chainalysis to host honest nodes that participate in the network 16:51:43 So this would be opt-in, with the embedded data? 16:51:45 indistinguishable from the reference software 16:52:23 Not all defenses are equally easy to circumvent. 16:59:51 Meeting in 1 hour - bring you own pet spynode :) 17:05:02 jpk68: this new entity is not confirmed to be linked to LionLink 17:05:51 Ah, good to know 18:00:02 Meeting time. Hello! https://github.com/monero-project/meta/issues/1468 18:00:14 Hello 18:00:25 Hello 18:01:57 Hi 18:02:00 Maybe more people will join a little later, but let us start already with the reports from last week 18:02:00 hi 18:02:30 waves 18:03:01 Howdy 18:03:02 will push an update to #9464 in a minute with some details in a comment 18:03:07 👋 18:03:21 Alright, people arrived :) 18:03:59 Today is the day for next stressnet? 18:04:39 @jberman:monero.social certainly knows details 18:04:59 yea! 18:05:48 FCMP++ fork is activated on beta stressnet v3, a little over 200 blocks until the second fork activates that stops allowing legacy txs entirely 18:06:10 @sneedlewoods_xmr:matrix.org: https://github.com/monero-project/monero/pull/9464#issuecomment-6000228481 18:06:10 (will check the comment after the meeting, may contains errors or bad formatting) 18:07:23 we also (finally) tagged v0.18.5.3, release notes can be seen here: https://github.com/monero-project/monero-site/pull/2710 18:07:27 release should be out tomorrow 18:07:51 this was the last release of v0.18 (unless there is a major regression) and the next release will be from v0.19 branch 18:07:56 no more c++14/c++17 split 18:08:00 v0.19* 18:08:15 👏 18:08:27 Quite a healthy list of contributors! 18:10:31 Quite a number of things look to me like things that AIs may find in code that nobody looked at for a quite a while. So we reduce technical debt, and also reduce the attack surface 18:11:37 For people who are maybe unaware, stressnet has its own dedicated matrix channel: https://matrix.to/#/#monero-stressnet:monero.social 18:14:03 I joined Monero a few months after RingCT went live, so I just missed much of the fun there. Won't happen again with FCMP++ :) 18:14:05 regarding the linenoise PR by jpk68, external/linenoise is upstream antirez/linenoise@a473823d plus about 515 changed lines in a single commit that doesn't record the base hash. shouldn't it commit a473823d untouched first and the changes on top, like for easylogging++ ? 18:14:46 I agree that this would be more ideal, though perhaps a better question would be whether it's worth it at this stage 18:15:40 What is "linenois"? 18:15:56 Er, linenoise of course 18:16:07 https://github.com/monero-project/monero/pull/11208 18:16:08 We are replacing the dependency on GNU Readline with a vendored and permissively licensed alternative 18:16:20 Ah, ok, that one 18:16:25 As a side-effect, this lets us remove something to the effect of 90K lines of code 18:16:33 *in dependencies 18:16:38 What? 18:16:47 Nice. 18:17:05 I'm dumb, nevermind, it's ~34,000 lines 18:17:26 I don't quite get this arbitrary loc argument, is linenoise battle tested like redline? Has someone reviewed all of linenoise code to make it worthwhile? 18:17:53 > used in Redis, MongoDB, Android and many other projects. 18:18:14 Well there's a start, but sounds like no one reviewed the code 18:18:20 yes it's been heavily reviewed tbh 18:18:44 well me and Ap4sh did 18:18:47 @vtnerd: Would you mind elaborating on what you mean by that? You mean the changes I made to the vendored code? 18:19:36 I mean if we're going with the smaller lib (perhaps not as small as I thought), has someone associated with Monero reviewed this code injection into our codebase 18:19:50 Like actually read linenoise source 18:19:54 Committing a473823d untouched first would let everyone see what changed vs upstream with git diff in monero's own history. Right now that means fetching antirez/linenoise, working out which commit it is and relying on his repo > <@jpk68:matrix.org> I agree that this would be more ideal, though perhaps a better question would be whether it's worth it at this stage 18:20:05 Hmm, the work done to fix all problems as detailed in the PR don't look exactly like an easy drop-in replacement at first sight? 18:20:08 @thomasbuilds:matrix.org, Ap4sh and @jpk68:matrix.org sounds like the new generation of monero maintainers 18:20:10 im feeling so old 18:20:15 @vtnerd: I assume tobtoht has done something like that, considering he suggested this change 18:20:50 @syntheticbird: wdym lol 18:21:59 @thomasbuilds:matrix.org: don't be naive or i'll consider you as an ORCHESTRA asset. ARE YOU CIA THOMAS ? JUST KEEP BUILDING AND TAKE CREDITS WHEN IT COMES TO YOU ALRIGHT 18:22:01 btw tobtoht supports the PR, this isn't a non wanted PR so far 18:22:10 Jeffro was also in support of the change, IIRC 18:22:10 We can't use readline because it is GPL, which is incompatible with BSD-3. 18:22:14 @jpk68:matrix.org: I think the answer is no, but I guess it's not terrible as it's not a niche as I thought 18:22:16 I could be misremembering something 18:22:21 my bad we're in meeting, sorry to yell like that 18:22:38 "requested that I work on it" Maybe better say tobtoht proposed that you work on it 18:22:44 Orchestra? 18:22:51 Ah, that is sadly a good reason @tobtoht:monero.social: 18:23:09 @syntheticbird: I still don't get it lol 18:23:21 like u say the PR is under reviewed still? 18:23:37 Is that directed at me? 18:24:02 @rbrunner7: I asked tobtoht for work to do because I was done everything else, and he said that would be a good thing to work on. Make of that what you will ;) 18:25:35 @thomasbuilds:matrix.org: Im just saying you three have been contributing a lot in the recent months and that is great to see, even more to see you are collaborating between each others and not isolated PR farms. 18:25:56 So, do I read the room correctly that nobody is really enthusiastic about this change, but it seems to have a reason with the license problem, and nobody intends to veto it as of now? 18:26:21 I'm certainly enthusiastic about getting rid of more bloated GNU stuff :D 18:26:31 Ok, that 18:26:42 @syntheticbird: ok yeah I get it basically I shouldn't assume it's "heavily reviewed" u mean 18:27:05 it appears one advantage is that it works under any TERM setting, currently our release binaries fail if someone uses a more exotic TERM setting 18:27:38 As a side effect we also get rid of ncurses. 18:28:14 What's to say about ncurses? It's big and only used for readline? 18:28:31 Yes and yes. 18:28:41 I see. 18:28:48 Also I can't figure out how to update it in depends. 18:29:13 It's cursed, clear as mud why you don't succeed. 18:30:08 Looks like we have reached something like "loose consensus" to indeed go ahead with this. 18:30:44 It would be good to hear opinions on this, as well ^ > <@thomasbuilds:matrix.org> regarding the linenoise PR by jpk68, external/linenoise is upstream antirez/linenoise@a473823d plus about 515 changed lines in a single commit that doesn't record the base hash. shouldn't it commit a473823d untouched first and the changes on top, like for easylogging++ ? 18:30:55 That is, the commit situation 18:31:16 You could split it into two commits? 18:31:38 That would be possible, yes, I just wanted to know if it's considered necessary 18:31:54 If everyone is fine with how it is now, I suppose it can just stay as it is 18:32:12 I don't particularly care much because I don't think we'll grab patches from upstream. 18:32:46 why do you exclude this? 18:34:37 About another subject: A while before this meeting there was a discussion here in this room about a new tool useful in our ongoing fight against "spy nodes" - if interested check the backlog and this PR: https://github.com/monero-project/monero/pull/11474 18:35:43 Ok, something else still left to discuss today? 18:36:24 @sneedlewoods_xmr:matrix.org: If anyone can take a look at and provide feedback for F3 or F4 for the first report in the linked comment, that would be appreciated 18:36:45 not needed now though 18:36:54 Will try to have a look 18:37:51 Thanks, already sent you a DM about another matter 18:38:31 I think I can close the meeting proper. Interesting stuff today! Thanks everybody for attending, read you again next week. 18:38:40 Thanks everyone, cu 18:38:47 This was certainly one of the meetings of all time 18:39:29 @thomasbuilds:matrix.org: anways main reason for me is for clean history of what happened 18:40:00 It seems no one is opposed, so I guess I can just do it 18:40:18 Will take six to eight business weeks to figure out how to fix the Git history /s 18:41:23 thanks it's a small thing to do with no downsides 18:42:11 afaik