Monero Dev Meeting Note Highlights – 2018-01-14
Logs
351 lines
rehrarHere's the agenda.
rehrarWe kinda did greetings, but if anyone else is present, please say "hi"
gingeropoloushi
jtgrassiehi!
pigeonshi!
Maxithihi
pebxhi
ferretinjapanhi
binaryFatehi
rehraralright, so let's get started then!
rehrar2. Brief review of what's been completed since the previous meeting
rehrarAnyone have anything exciting or fun to report since our last meeting a month ago?
rehrarGet any fun gifts for the holidays maybe?
Maxithi24 commits 4 days ago.
MaxithiOn the main core. Nothing special. Maybe a little cheering for cslashm and hist first Ledger PR.
endogenicwell we've made some progress with apple, fyi … confirmed there was an account mixup on their side with a human being and we're finally getting that sorted. will update you all as i hear things
rehrarthat IS pretty exciting Maxithi
endogenichave to submit some docs, etc
hycnothing much to report here
sgphi
rehrarendogenic dude, that's gotta be thrilling to get some news finally
rehrarArticMine, do you want to speak briefly about Kiev?
endogenicrehrar yeah i mean these things usually happen instantly so once i called them they were like yeah that's no good let'sget that sorted
ArticMineSure, but it is not really dev related
rehrarjust a brief four sentence thing should be alright
vtnerdafter a discussion with luigi1111, I should be proposing an optimization for wallet scanning that mymonero is using - want to test with a standard wallet to give real world perf numbers before PR
vtnerdah sorry ArticMine
rehrarno, go ahead vtnerd :) anyone and everyone with an update feel free to let us know
vtnerdok, this optimization is separate from the ASM work in monero-project/supercop (still pending a PR), so it works with standard ref10 (i.e. all platforms)
vtnerdit will require a few changes to src/crypto.{h|c} so be on the lookout for that PR
gingeropolousis this just for your standard refresh? or related to remote viewkey scanning… or both?
vtnerdthis is specifically for scanning receiving outputs on a primary address
vtnerdit does not benefit subaddresses, but does benefit simplewallet / GUI / anyone doing viewkey scanning
medusalets talk about the elephant in the room ? bulletproof is in all varieties on testnet
rehrarAlright. Thanks. :) Anyone else? If not we'll move on.
rehrarYe, that's next.
moneromoooThings of note since last time: build hardening (currently breaks static builds though), DNSSEC fix on windows, DB and sync hang fixes, secure memory wiping (wip), misc code review and fixes, updates now use HTTPS and can resume downloads
ArticMineIn brief I attended BTC awards CIS and participated in the coins panel. There were two other coins present Dash, ZenCash. Overall it went very well and provided exposure to Monero among the CIS (Commonwealth of Independent state) crypto communities
endogenicmedusa: last i heard MRL was looking for people to review and audit bulletproofs
endogenicsarang would be able to give you more information
ArticMineThat is my brief update
ArticMineWhan it comes to bullteproof this has to move to September
endogenicmoneromooo: beast mode
rehrarArticMine, care to elaborate?
moneromoooNot all of this by me fwiw.
hycInteresting ArticMine - CIS already has monero experience, thru the Karbo clone in Ukraine
medusaGUI is merging sub addresses finally, dsc is working on the black theme sucessfully and we hope to go live in march with 2 progress bars instead of 1 (sync and refresh separated). also we added a "change password" feature to the GUI
ArticMineFP spoke with Pieter Wuille who is recommending September
sarangJumping in for any questions about BPs that I can answer
ArticMineIs there a plan for further review?
ArticMineof B
sarangYes
ferretinjapanAre there any showstoppers on bullteproofs that would justify moving back rolling it out?
sarangInitiating contact w/ an audit group that fluffypony knows
sarangIt's not fully clear what their scope of review can be, but I will find out shortly on a phone meeting with them
sarangThe goal being additional third-party review
hycferretinjapan: there's no known problems. additional review is being planned to avoid unknown problems
rehrarYeah, so we naturally moved onto third item on the agenda, which is code freeze, March hard fork. Bulletproof is the biggest one. Everyone's thoughts?
binaryFatebeyond code review, sarang what is your opinion on the need of review of the math itself? I'm worried the paper has little time to go through peer-review. But maybe match checks out so obviously that it's not needed.
sarangThe math is solid. Additional peer review is always a welcome thing, though with diminishing returns over time
binaryFate(talking about the BP paper – not the Monero specific implementation)
sarangyes
sarangIn the meantime, moneromooo and I have been working on some additional optimizations for proof verification that are independent of the math or specific proof implementation
ferretinjapanI guess thats the thing, how much "value adding" is an additional audit for now justified?
binaryFatenice
sarangWe've had a select number of informal reviews
sarangbut no formal third-party review yet
ArticMine… but still this warrants a delay to September for BP
sarangTo get the review done? Almost certainly
sarangThe advantage to the delay is that we could bypass single-output BPs altogether
sarangfigure out the multi-output fee structure, and release that
medusamulti output for march is a non option ?
sarangCorrect
ferretinjapanseems incredibly wasteful :/
sarangIt requires a change to the way fees are computed
hycI'm in favor of skipping single-output. would be a lot of work for a very short-lived feature.
sarangYes, though single-output proofs will verify with the multi-output code with no changes
ferretinjapanyou'd be better served making sure that there is a fallback to BP instead of pushing back entirely.
rbrunnerIsn't the work almost all done anyway?
ferretinjapanthis is a classic case of perfect being the enemy of good.
endogenicrealized we didnt ping anonimal
sarangWe have multi code working now
sarangit's not in testnet yet
sarangand we don't have consensus about the fee structure
hyc^^
binaryFateI lean towards satefy in general. The ringct bug was close enough
Maxithihyc Same, better safe than sorry.
iDunkAgreed.
dEBRUYNEsarang: have you considered contacting the paper authors for review?
gingeropolousferretinjapan, the push to sept is about additional code review, not really about multiout
dEBRUYNE<binaryFate> I lean towards satefy in general. The ringct bug was close enough <= I can't survive another mini heart-attack probably
sarangWe've had informal review from one author, but the other authors are working on other implementations and I would not rely on them for a speedy review
hycI suggested soliciting reviews on the Cryptography email list
endogenicnice idea
hyccould also solicit reviews on twitter
dEBRUYNEBut presumably they would have some time somewhere between now and August?
endogenicthat too
dEBRUYNE\^ sara
dEBRUYNEsarang*
saranghyc: are you ok with making such a solicitation?
sarangI could certainly try the other paper authors just in case
moneromooobypass single-output BPs is not an advantage.
hycYes, can do. we need to decide if we'll offer a payment
sarangthough they are much more of a long shot
moneromooobypassing
ferretinjapangingeropolous, yes, but noones really given a definitive reason to delay except for an overabundance of caution, and by then it'll be unnecessary to even release. :/
endogenicis there a blurb/call-to-actionabout the request to review which i can pass to my relevant mailing lists?
hycmoneromooo: waiting till sept means we can settle on the fees without rushing
saranghyc: fluffypony indicated we could make funds available for the review, but you'll have to ask him for the details
moneromoooIt's not a lot of work, since it exists now.
ArticMineWe are only delaying single output not multi output
moneromoooMulti output bulletproofs are in C++, but not plugged into the consensus, which is the annoying part.
sarangAnyone interested in doing a review and wanting details can contact me at [email protected]
endogenicsuch as indicating where to look in the code or repo, background docs, review process, submission format, etc?
hycendogenic I'll start drafting an email announcenemt
sarangI'll contact the other paper authors separately
sarangand will still press on with the audit group to see what their scope is
luigi1111wthere's not much work saved by skipping single output BP
moneromooowaiting till sept does not change anything for fees, since multi out is september anyway.
luigi1111wbut the review time is too close for my comfort for march
luigi1111wI think that's the general thought anyway though
hycalternatively we could delay the March release into April
sarangfluffypony was very opposed to that idea hyc
hycok
ferretinjapancould always push back the march hard fork a month or two like last time for RCT.
sarangnot saying that's the deciding factor though
luigi1111wrct was pushed forward actually
rbrunnerWhat's fp's main argument?
ferretinjapanoops, I c.
sarangHe thought that regular releases were an advantage
rehrarhe says that people are getting used to the hard fork schedule
sarangdue to predictability
endogenicrbrunner: for?
ferretinjapanrbrunner, probably his snity ;)
ferretinjapan*sanity
rbrunnerAgainst the 1 month delay
luigi1111wmy own shed says 3 months with fork in June :D
endogenicbc it c hanges the hard fork periodicity etc
endogenicguessing
rehrarI think six months is not super long to wait for bulletproofs. It'd be another thing entirely if we were discussing putting it off for a year or so, but not much is lost by being a bit safe and waiting.
gingeropolous^^
ferretinjapan6 months is a looong time in crypto
luigi1111wdid someone mention future pruning
luigi1111wruns
sarangsigh
gingeropolousfuture pruning? you mean double blob?
endogenicferretinjapan: better than there being no crypto bc we broke it
gingeropolousruns
sarangPeople also want Lower Fees Now
sarang(tm)
vtnerdwe all want ponies sarang
sarangor at least, a vocal minority does
dEBRUYNE<rehrar> he says that people are getting used to the hard fork schedule <= I doubt most of the economically sensitive nodes are aware of our schedule
ferretinjapanendogenic, touche, but there comes a point wher you gain nothing by waiting.
ErCiccioneI think in this phase keeping the scheduled 6 months shouldn't be a priority. Agree with dEBRUYNE
rehrarIf we were to push back by two months from March to May, would that be a good compromise in people's opinions?
luigi1111wno, but 3 months from March to June…..
sarangIf we really want additional review, the timeline will need to be flexible to the reviewers
gingeropolousbut not rushed
MaxithiI'd rather have 3 months than 6.
sarangacademic review can be a frustratingly slow and unpredictable process
rehrarok, from March to June then
endogenici agree with sarang, we cant dictate when third parties deliver
iDunkThis just looks like a repeat of the last meeting.
sarangbasically
luigi1111wiDunk :)
rehrarit basically is :P
endogenicwe also dont have enough info to know how long sufficient reviews would take at this stage
endogenicso not sure it's a good idea to judge things yet
ferretinjapanrehrar, I think so, but I think it's important to be realistic, rushing is bad I'll admit (we don't want stressed devs!), but unnecessarily putting it off is also bad
dEBRUYNE8 months seems like ample time though
rehrarhard part of being decentralized with no leader is that consensus on stuff is hard
ArticMineWhy not go ahead with March and if BP is ready then to an early September HF? for BP
endogenicoh yeah what about the meeting agenda?
luigi1111wwell there's really nothing left for march
sarangMaybe best to wait to hear about potential reviewer interest
sarangotherwise this is all moot
luigi1111wbut we can fork anyway to keep schedule
saranga totally moo point
rehrarhow long do we wait though, since code freeze is soon?
sgpWhy not plan a hardfork a month after the BP review is ready, whenever that is?
endogenicmooot*
ArticMineMultisig is in March
luigi1111wnot forking
rehrarMultisig doesn't require a hard fork though
dEBRUYNEMultisig doesn't require a hard fork though
luigi1111wbut in release yes
dEBRUYNElol
luigi1111wyou guys
rehrarlel dEBRUYNE
endogenicwhat the
rehrarof one mind
gingeropolousbut still could provide bugs to squash
endogenicjust got the heebies
dEBRUYNEbots rekt
pebxso we don't need any hard fork in march without bp?
ferretinjapansgp, I think thats a solid idea, but "consistency" may be a factor in that.
luigi1111wpebx part of the idea of scheduled forks is to do them anyway
rehrarcan we go through the rest of the meeting and come back to this topic?
sgpI think the rewards are so large it's worth considering breaking consistency
rehrarjust so we make sure everything else gets done?
rehrarthen we can discuss till the end of time at the end
luigi1111wor 1hr, whichever is less
endogenicrehrar: what are you, some kinda pragmatist?
ferretinjapanluigi1111w, noone said anything about FIXED schedules ;)
gingeropoloussgp, indeed, that was the logic begind the earlier than planned ringct fork
rehrarSome people gotta go or don't want more than an hour
luigi1111wlol
dEBRUYNEWhat else is there to discuss though
endogenicdEBRUYNE: how about what kind of soda you're going to get for rehrar?
iDunkSubaddresses and fees for the next fork.
rehrarshrug screw the agenda then. We're not beholden to it right. ;) Just trying to think of people who might want to discuss code or issues.
ArticMineI do not see a fee change without BP
pebxi'm not a mathematician, however i would prefer single output bp in march/april/may than waiting for multi output with the even more controversial change in fee structure
iDunkEven if the value has gone up ~40x since the last adjustment ?
endogenicrehrar: dont be silly, no one wants to discuss issues… :P
rehrartroo, ok BP it is
sarangpebx: the fee change shouldn't be that controversial
luigi1111wArticMine smooth had some arguments for reducing fee
sarangthe alternative is DOS
rehrarHow soon should we be looking to come to a decision? This meeting? Next?
rbrunnerWhat is DOS?
Maxithiendogenic: rehrar: dont be silly, no one wants to discuss issues… :P <= I do, please label them so I don't have to make lists and have dEBRUYNE tag 150 of them at once ;)
endogenicrbrunner: denial of service?
rbrunnerRegarding fees?
ArticMineWE can lower the min fee but people are not using it anyway
endogenici.e. spam?
serhackhi
dEBRUYNEFor the current range proofs you'd need to tweak the penalty
luigi1111wnot gonna happen
luigi1111wby march
ArticMineSpam to the txpool is a risk if we go too low on the min fee
dEBRUYNEWhat's not going to happen by march luigi1111w?
luigi1111wtweaking the pentaly
pebx@ArticMine most users don't know about a minimum fee…
luigi1111wpenalty
pebxI see this all day in the Telegram groups "why is the fee so high?"
ArticMineOne option change the wallet default to the min fee
dEBRUYNEThe per kB fee isn't even that high fwiw
pebxmost services also use the default… like mymonero does
dEBRUYNEThe transactions are just obscenely large
endogenicpebx: dont worry, i'm getting close to adding priority setting to new apps
pebxdEBURNE isn't that why bp is so important?
dEBRUYNERight, but we shouldn't rush it
Maxithi@ArticMine: One option change the wallet default to the min fee <= I actually consider that a good idea as no one uses that fee…
ErCiccione^^
dEBRUYNEThat's going to result in a clogged mempool if we want to expand
luigi1111wI use that fee
moneromoooThen noone would use the current fee once it's no longer the default.
iDunkSo do I.
luigi1111wwell I use both kinda depending
jtgrassieBP shoudnt be rushed but equally shouldnt be held back to a schedule
luigi1111wdEBRUYNE some of that is unavoidable
MaxithiYeah but I mean none of the people here do used it. That's the issue
luigi1111wtwo opposing forces at work
rbrunnerIt's nice to have a fee level below default to be able to say "I can wait"
dEBRUYNEluigi1111w: According to smooth miners will just pick the default or higher fees and leave everything else in the mempool
dEBRUYNEBecause they can only use the last transaction for the penalty
luigi1111wwell I'm pretty sure that was originally according to me :)
dEBRUYNESo you can't have 23 min fee txes and then one default
binaryFate<ArticMine> One option change the wallet default to the min fee <– +1
endogenicisn't that a bad idea? \^
moneromoooIt is.
luigi1111wit's an idea
luigi1111wI'm not sure if it's bad or good
moneromoooUnless people don't use monero much for half a year.
endogenichm it would be cool to model this
Maxithigot to go, see you guys
endogenicbut Maxithi you didnt get to chat issues
luigi1111wsorry I got lunch gotta run too
luigi1111wgrumbles about meeting time, again
ArticMineIt is the human herd mentality
Maxithiendogenic Yeah never mind, just wanted to ask people to label more issues to make it easier to know which one require assistance.
Maxithiagrees with luigi1111w
dEBRUYNEWhat if the wallet changes the level dependent on the backlog?
endogenicMaxithi: maybe you could make an issue about that?
dEBRUYNESo it'd use min fee if mempool is < 1000 kB and default if > 1000 kB?
dEBRUYNE\^ just using arbitrary numbers
rbrunnerWouldn't that just result in a lot of angry people, like "Why the hell I paid so much"?
rbrunnerIf you don't make it very transparent
ArticMinedEBRUYNE What if the wallet changes the level dependent on the backlog? <— That would work
dEBRUYNErbrunner: How so?
hycendogenic, sarang: email draft https://pastebin.mozilla.org/9075994
dEBRUYNEYou can add a note in the wallet
pebxisn't that even more confusing for users?
dEBRUYNE"changing fee level because of backlog"
rbrunnerI am afraid people would not understand the mechanism
rbrunnercompare fees, see that they are different for different people
rbrunneror different times
rbrunnerand shout "scam" or so …
ArticMineIt is what Bitcoin does
endogenichyc: term "output" used before definition
dEBRUYNEThat seems a bit too far stretched
rbrunnerWell, people now even don't "understand" they can go one level lower for the fee, right?
saranghyc: I didn't see any Java code from the authors. I wrote our Java proto code
rbrunnerMany of them, anyway
DaveyJonesrbunner i don't think so … most people arn't even schooled on default vs 0.25x fee
saranghyc: line 5, I'd say "are confident" instead of "are pretty confident"
hycsarang: ok
endogenichyc: maybe capitalize 'Java' and add a 'Therefore, ' before "The Monero Project"
rbrunnerYes, how will they understand how that works if it get more complicated?
rbrunnerThat's my fear, basically
endogenicoops
endogenic"mainnet. The Monero Project"
rehrarI'm torn tbh. I see both sides.
pebx@ArticMine what bitcoin does is one of the major reasons, why people are looking into Monero… fees are one thing the average bitcoin user is always confused about
DaveyJonesbasically they don't need to understand most of people moaning about tax fee's are people just count the $ value of the fee… not how it was achieved
endogenicotherwise looks great, tyvm!
DaveyJonesso they basically dont care about the mechanism but the price
DaveyJonesand wasnt the proposed mempool fee sth that should happen automatic? or did i misunderstand that
ArticMinepebx Bitcoin has fundamental fee problems that have nothing to do with wallet choices.
saranghyc: maybe state that we "recognize the value of independent third-party review" to assure other readers that we're doing this as a "belt and suspenders" approach, rather than a legitimate worry about correctness
binaryFatepebx Bitcoin is a clusterfuck of floating fees and fees competitiong for a fixed block size. Here it's just about choosing the default among the 4 different levels, that's it. Then flex block cap takes care of the rest so you don't have your transaction in mempool for 14 days
sarangI don't want some silly media report that we don't trust our own code
jtgrassieIMO default fees should derived from mempool size and be completely transparent to user
endogenicbut sarang mass media never selectively reports
sarang-_-
endogenicthat wouldn't be true journalism!
ArticMineIt can be done at he wallet level.
binaryFateIt's just silly that when there is factually no backlog in the mempool, the fee does not default to low
rbrunnerCan the wallet query tx pool size already?
endogenicbinaryFate: simplewallet does warn at least
rbrunnerTechnically, is there a call?
endogenic"n block backlog at that fee"
endogenicoh, next meeting when? 2 wks?
sarangendogenic: I want to make it clear that we're taking a careful but confident approach that uses peer review, just as any good academic work does
rehrarIt was suggested earlier that if single output BP's are not merged in March, that the fee should be adjusted. If we are saying that we don't want to adjust the fee in March, I think single output BPs should be merged
endogenicsarang: i completely agree
jtgrassie13:02 < endogenic> oh, next meeting when? 2 wks?
rehrarYe, sorry. If anyone needs to leave. Next meeting two weeks.
rehrar28th
binaryFateThe warn makes people think "mmm should I use more fees", not the opposite. Who when using normal and seeing 0/1 block backlog cancels, and come back to use low fees? Nobody I guess, should be automatic maybe with just a y/n confirmation
saranghyc endogenic: move discussion of BP review to #monero-research-lab after meeting?
endogenicyes good point
endogenicbinaryFate
hycok
endogenicthough i dont know about automatic, from a ux standpoint, but thats just me
rbrunnerAutomatic always takes away some choice, right?
endogenicnot necessarily
endogenicif the result is displayed clearly so the user can confirm before implicitly agreeing
endogenicthen it can be ok
endogenicquestion is
endogenichow do you do that in a CLI?
binaryFateWhat we're discussing is just selecting the default, not forcing the user
endogenicmay end up needing some kind of prompt or to enhance it to update some kind of display in-place
endogenicyes
endogenicmust be off, myself
endogenico/
rehrarCan we maybe plan an "urgent" meeting sometime this week or next weekend?
rehrarJust to discuss BP's and make a decision?
endogenicrehrar i'll attend any meetings
rehrarOr is that not a good idea.
sarangrehrar: Do we want to wait on review feedback?
rbrunnerIt seems BP might merit that, a special meeting
sarangTo at least see who if anyone will review?
endogenicsarang: personally i would
rehrarSarang, do we have an ETA on that?
DaveyJonesBP + fee talk ?
ArticMineI am fine with that say next Sunday
DaveyJonesi guess both run a little short today
endogenicor else i would hold the meeting with the primary objective of resolving how we can solicit and accomplish the review
DaveyJonesand need more thoughts
sarangrehrar: I have a phone meeting w/ the audit group on Tuesday
rehrarPerfect.
sarangI assume hyc can post the Call For Reviews after we finish proofreading it
rehrarI can make an issue depending on that.
sarangand I'll contact the other paper authors today
rehrarI will put a "tentative" next meeting for next Sunday
sarangroger
rbrunnerSounds good
rehrarAnd if anything changes, I'll change it.
sarangI'll post any updates on the research IRC as I hear back
rehrarYou can comment on the GitHub issue too plz.
sarangsure thing
rbrunnerI am sure companies would arrange meetings of whole days for something of the importance of BP
rbrunnerJust saying :)
rbrunnerJust to reach consensus
rehrarGood idea rbrunner. 4 hour meeting planned. Followed by a small snack break. Then another four hours.
rbrunner:)
ErCiccionecan we talk about code freeze now? I really need to know the date for both the localization team and the GUI guide that will go in the binaries
rehrarHarder to discuss now that lots of people are gone unfortunately. :/
rehrarBPs took most of the time.
rbrunnerWith breakthrough still somewheat elusive, if you ask me …
rbrunnerBut maybe that's a good thing.
ErCiccionerehrar: yep..but that mess things up a bit for me. still around mid February btw, right?
rehrarrbrunner, shows we're actually decentralized?
rehrarDunno erciccione, I think it was supposed to be January, but don't quote me.
rbrunnerrehrar: Yes.
sgpiirc it's supposed to be 2 months
sgpBefore the hf
rbrunnerWas one of the important points of fp's speech in Zurich that I could attend. Impressed folks, I think
ErCiccionererhrar: it was, but last time i asked dEBRUYNE he said more or less around February. i guess i'll just wait to know more
binaryFaterbrunner what point?
sgpI think it's a good idea
rbrunnerThe good decentralization of Monero
binaryFateErCiccione as long as hard fork date still somewhat open and discussed, I think it's fair to assume code freeze will not happen soon. That's probably the most you can get now.
sgpI think some localization stuff is low-risk though and could be merged a bit later as long as the binaries can be built with sufficient time. Not my deceion though
BillyI do think we need to resolve the BP issue ASAP. It needs a special meeting because these general meetings aren't going to do it.
rbrunnerYou could say that after two regular meetings, yes
rbrunner1 hour is damn short, even if we don't have to emulate endless corporate meetings
ErCiccioneok, thanks binaryfate sgp the problem is that some translator keep asking that because they need to organize their time, but they don't want their work to be left out from the official release
Sources and notes
- Meeting log: Logs for the Dev Meeting Held on 2018-01-14, getmonero.org. The official post carried the raw log and pointed readers to Hello Monero for the overview.
- Original Hello Monero URL slug recovered from the Wayback Machine index; the original page content did not survive, so this reconstruction carries the log as archived by the Monero Project.