Перейти к публикации

Solana


cp287

Рекомендованные сообщения

Round 1 SAFT: 30/03/2018

Round 2 SAFT: 01/06/2018

Round 3 PETSA: 09/07/2019 - 09/09/2019

Round 4 SAFT: 28/02/2020, $2.4M @ 25c

Round 5 Coinlist Auction, 24/03/2020, $1.6M @ 22c

photo_2020-02-29_20-09-35.thumb.jpg.72cb3ddcce98864e70d0b40e56452054.jpg

 

Таблица участия в тестнетах: https://docs.google.com/spreadsheets/d/1pEdQoSxBakwsaHuGgiQVoYcu49WWbZfhOus_H5ZyDL4/edit

Официальный сайт: http://solana.com/

Мониторинг статистики сети: https://solanabeach.io/

 

Ссылка на сообщение
  • Ответы 69
  • Создано
  • Последний ответ

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

Кин-у понравилось мигрировать, а теперь за это еще и платят 😁 https://www.coindesk.com/fed-up-with-its-fork-of-stellar-kin-is-looking-to-move-onto-solana   The Solana Foundation would ac

Алевтина о стратегии команды сделать масштабный разлок: источник: tg -Solana RUS

Solana в гостях у Near Protocol (на английском):  

Posted Images

Подкасты с Анатолием Яковенко, CEO Solana.

 

Incrypted + наш форум: 

 

Forklog: 

 

С нуля до нуля по крипте: 

https://t.me/nullpodcast/12

таймкод: 

2:30 - Анатолий рассказывает, как он попал в блокчейн индустрию
7:00 - Анатолий и его мнение по эфиру (в том числе эфиру 2.0) 
7:30 - куда стоит двигаться нашей индустрии?
8:40 - что такое moore's law?
13:37 - объяснение принципов работы Соланы на пример радио 
16:20 - разная классификация проблем, которые решают Солана и Эфир 2.0
19:30 - кто работает над Соланой?
23:00 - про интеграцию EVM Либры
31:40 - про валидаторов в Солане 
38:00 - Солана b2b или b2c проект? Кто целевая аудитория блокчейна?
42:30 - зачем Анатолий дает интервью на русском
44:10 - Дальше Только Больше

 

Ссылка на сообщение

Подкасты с Анатолием Яковенко, CEO Solana.

 

Incrypted + наш форум: 

 

Forklog: 

 

С нуля до нуля по крипте: 

https://t.me/nullpodcast/12

таймкод: 

2:30 - Анатолий рассказывает, как он попал в блокчейн индустрию
7:00 - Анатолий и его мнение по эфиру (в том числе эфиру 2.0) 
7:30 - куда стоит двигаться нашей индустрии?
8:40 - что такое moore's law?
13:37 - объяснение принципов работы Соланы на пример радио 
16:20 - разная классификация проблем, которые решают Солана и Эфир 2.0
19:30 - кто работает над Соланой?
23:00 - про интеграцию EVM Либры
31:40 - про валидаторов в Солане 
38:00 - Солана b2b или b2c проект? Кто целевая аудитория блокчейна?
42:30 - зачем Анатолий дает интервью на русском
44:10 - Дальше Только Больше

 

Ссылка на сообщение

Запускают аукцион на Coinlist с возвратом 90% цены для желающих. Возможность контролировать цену будет обеспечена продажей токенов  (на каждый застейканный токен могут продать один токен) в случае, если цена будет выше цены аукциона, т.е. пампа ждать не приходится.

На аукционе будет продано 8М токенов.

Solana Staked Price Guarantee-49c16ed7cef09a9818a21ccfbac4638daa3292c40910e044647e55bf2364b08c.pdf

Ссылка на сообщение

Собранная основная инфа о сейле от ICO Analytics:

 

CoinList will host Solana (SOL) auction as of March 17. The dutch auction will have a start price of $4 USD and the bottom price of $0.02 USD without purchase size restrictions. Solana will offer 90% guarantee of the auction price for 12 months period for those who will stake the coins during 3 month from the auction. 90% guarantee of the auction price will apply to coins purchased via exchange as well. Solana will be selling additional portions of coins from Community reserve pool(34% of total supply) based on the formula 100 SOL staked=100 SOL to be sold if the market price will be above the auction price. 

Sale details:
Total supply(Genesis block): 500M SOL
To be sold on CoinList: 1.6% (8M SOL)
Raised during private sale: $23.9M USD
Private sale prices: $0.04 - $0.25 USD
Sold during private sale: 35.6% (178M SOL)
Private sale lockup: 9 months 
Team coins lockup: 9 months

 

photo_2020-03-11_17-30-02.thumb.jpg.e407efd90e5d5eb62e475c6bf9eb35f6.jpg

Ссылка на сообщение

Несколько нового контента от Solana.

1. Подкаст Solana под названием No Sharding: https://podcast.solana.com/

2. Серия постов для валидаторов: https://medium.com/solana-labs/validator-feature-series-6f1c479d70b3

3. Предложение для более надежной инфраструктуры рынка от CEO Solana Анатолия Яковенко (en): https://medium.com/solana-labs/elastic-circuit-breakers-9229097e6acf

Ссылка на сообщение

Вероятно, что-то изменят - или итоговую цену, или кол-во проданных токенов. На 0.24 при традиционной на таких аукционах "облаве" залагал сам сайт коинлиста и часть заявок юзеров не исполнилось, хотя аукцион якобы закрылся ниже (на 0.22).

Я обв за увеличение кол-ва токенов 🙂

Ссылка на сообщение
14 минут назад, Neftyanik сказал:

У меня даже по 0.26 не исполнился.

Думаю, ты видел, но на всякий копирну

"У всех, у кого не сработали ставки выше цены закрытия аукциона, пожалуйста, напишите об этом email на адрес team@coinlist.co с копией raj@solana.com. Мы работаем над решением этого вопроса."

Ссылка на сообщение
50 минут назад, T сказал:

Думаю, ты видел, но на всякий копирну

"У всех, у кого не сработали ставки выше цены закрытия аукциона, пожалуйста, напишите об этом email на адрес team@coinlist.co с копией raj@solana.com. Мы работаем над решением этого вопроса."

Да, уже ответили, чтобы я подтвердил свою ставку еще раз. Но в письме не указанно за счет чего они ее восстановят. Просто написано про внутреннюю ошибку платформы.

Ссылка на сообщение

Прислали инвесторский отчёт. Мечтают о 5 миллионов активных пользователей к концу года (WHAT?), и почему-то присылают остальные обновления в виде deck-а (отдельного пдф😞

Цитата

March has probably been the most eventful month of any of our lives. The macro environment created a lot of uncertainty for Solana, but we feel we have navigated it and will come out of this crisis in a position of strength. We’ve increased our runway to accommodate the team achieving every major goal we set out to achieve when we started Solana. We’re expecting to end the year with at least 5m monthly active users, and we’re shooting for 10m. The latest auction results give us confidence we have the marketing engine and community support to weather any economic climate. We are oddly more confident than we’ve ever been. We've attached a deck outlining the highs, lows, and details of this month. It is packed with information.

We’re driving toward our first listing, and expect it to go well. We’re the only single-layer blockchain capable of supporting NASDAQ-level throughput, and we’ve outflanked all the other high-throughput chains in launching. This excellence in execution and community building is finally being recognized. We’re having conversations every week with the largest, most well-adopted projects in the space. We are transitioning into more of a sales mentality, aiming to win the best teams and products over in 2020 and beyond.

We could not have done any of this without your support. In 2020 we will have a few major asks, and we are very grateful for your response to these calls to action. Please always feel free to schedule a call with us if you have any questions or concerns. Onward and upward!

 

Solana Investor Update - March 2020.pdf

Ссылка на сообщение

5 млн 🙂 угу.

 

Аукционщики до 7го апреля должны создать кошелек и заполнить форму. На вопрос "Когда токены можно будет перемещать?" последовал ответ "С большой долей вероятности 7 апреля)". Теперь надежда на то, что в этот же день будет офиц листинг, иначе опять всякие хотбиты/билакси убьют эффект новичка (см кадена)

Ссылка на сообщение

20-минутная запись сегодняшнего звонка Валидаторов: https://zoom.us/rec/share/w9xtBLfrpl9IHoX8tBzdd7Z6Btrnaaa82ylKrKULyh3hqcuETPVfvNyts6vy9QMc

Обсуждается прошедший аукцион, третий этап Тур де СОЛ (спойлер - он откладывается до решения некоторых технических проблем), переход Соланы из юрисдикции США в юрисдикцию Швейцарии и нафига козе баян зачем был создан механизм гарантии цены.

Ссылка на сообщение

Битва проектов. Лидер проекта Solana Анатолий Яковенко и основатель Ava Labs Emin Gun Sirer зацепились языками в одном из телеграм чатов по теме "какой блокчейн самый быстрый":

Цитата

Anatoly | Solana: hi folks! on the public network we have sustained 20ktps over 20 min, over 65k 5 second peaks, blocks are 400ms. Whats neat, is we have our own flavor of asynchronous pbft implemented, so number of nodes that can participate in consensus is limited by the TPS, not the other way around :).  you can take a look at our validators humming along here https://solanabeach.io/#/ I thought it was to make fun of crypto.  Making blocks has product market fit.  We can make them bigger and faster than anyone else 🙂

Emin Gün Sirer: Jane makes 1 decision per year. But when she does, she bundles a gazillion transactions together, so she decides on a gazillion tx's at once. So now, does Jane get to claim gazillion/year = millions of transactions per second?

A: Depends if those transactions have state dependencies or not

E: This is literally the oldest trick in the I-want-to-get-tenure-but-have-a-slow-system handbook. I hope practitioners are wiser. No, it really doesn't.

A: Yes it does

E: Report decisions per second Anatoly. You can do it!

A: Nope.  Trades per second executed

E: Then all your numbers are meaningless.

A:Janes black box censorship resistant infailable aggregator is better then your block chain 🙂

E: I decided on a merkle root over a billion tx's, yay, billion tps!

A: All I care about is price updates per second on a dex. Each update has capital risked, each one can be captured. If you can make an aggregator that still works in that usecase, you won

E: These tps numbers are meaningless.

A: Nope :). Price updates per second is not meaningless. If you can do that in a sharded system. You won as well. I can’t take action on price changes between shards or rollups

E: Don't deflect by throwing in random red herrings. These numbers are disingenuous.

A: Yet, work in a dex that can evaluate a market with 30k price updates per second...

E: I don't hear you saying that you aren't using the oldest trick in the book to fluff up benchmark numbers. You're just defending the trick. SMH.

A: Yea. That’s just 2 hardware revs. We are cpu core bound

E: You should be I/O bound. These are all red flags Anatoly.

A: IO is horizontally scalable. It is. State replication can be done in just 2*X bandwidth. Imagine postgress, handling 2 million writes per second and propagating the log via BitTorrent.

E: That's not safe

A: No with 50:50 erasure coding

E: Still not safe. Did you reinvent primary-backup replication?

A: Ok. Great.  We got the not safe case to work 99.9% of the time

E: That requires trust in the primary

A: Rotate the primary

E: Still broken

A: How?

E: Are you live?

A: That’s how tdma works. We have 4 billion devices working.

E: Is Solana live, that

A: Yea

E: Great. TDMA is not BFT.

A: It’s not a database. It’s a block producers scheduler

E: You seem to have coupled the most minimal crash-but-not-byzantine fault tolerant system with some of the oldest tricks in the book to fluff up numbers. That's like EOS. But without signature accrual

A: You are making up fake goal
E: So, not BFT

A: It is

E: Safety and liveness aren't fake

A: Can handle up to 1/3 failures. Max outage is 1 epoch though. It’s at 2 days atm.

E: So, I get chosen as the leader/primary, I push a different update to every replica, then I go down. What's the state of your blockchain?

A: its async pbft. on top of tdma

E: 👆

A: next producer, if they have your block attaches their block to yours. if not, attaches to the previous block. if the rest of the chain has your block, they accept the new one, or fail it

by fail, do the same, either attach to it or not

E: Not safe

A: under what conditions? we have a reliable source of time. a vdf. yea, thats fair :), seems to work for now

E: VDFs do not accomplish most of the tasks that people think they accomplish.

A: they do force a delay quite well

E: A delay, sure, but not a fair delay. Everyone almost always wants the latter.

A: the only usecase that matters, is a DEX, how much can a single global market settle per second. how many price updates can i take action on. doesn't matter what tricks you use, if you can make it work, they are fair

E: You do not have the D in DEX

A: we can do, 30k price updates per second, that get some finality every 400ms

E: Did you implement the full PBFT, with view changes?

A: its asynchronous. but yes, view changes

E: Did you implement the view change protocol portion of PBFT? "views change" isn't the right answer, given how many corners seem to have been cut

A: easy to fud without a concrete attack

E: This isn't FUD, just a simple yes or no question

A: yes, there are view changes, there is a proposed view on every block

E: That's a clear "no". The view change protocol is the most critical part of PBFT. It's clear that your protocol isn't BFT. You built primary-backup replication. Ok great. Not safe

A: ok, great, its safe up to 1/3 attacking nodes

E: Can't be true if you do not have the critical parts of PBFT implemented

A: we do. we just dont synchronize on a single view. its synchronizing over 1 or more

E: Wanna know why people are saying what you've got isn't safe?

A: we simulate big and small network partitions continuosly

E: I don't think it's probabilistic

A: yea, its not. its fully deterministic

E: I think it's just a simple Deterministic replication protocol

A: yes, ava is just a dag 🙂

E: The real question is, if it's not Byzantine fault tolerant, why is it so slow :-).

A: its bft. the main change we made is evaluating over N views instead of just 1, which is the big slow synchronization point

E: Sorry Anatoly, you know I love you to bits, but at this point, it's fair to ask for proof or retraction

A: school is out 🙂

E: Alrighty.

A: i dont get paid to teach.  once you guys are live we can write a benchmark and compare

E: It's apples and oranges.

A: yes, correct. we have the same guarantee

E: We'd compare Solana to Postgres. Can you provide a proof? Of just safety

A: some kids who are getting their phd can, why would i spend time on proofs?

E: Because that's how you discover errors in your own claims. Lamport has a nice quote in one of his early papers about how so many people came up with purported consensus algos that are buggy, and that all protocols without a proof should be considered broken

A: sure, all whitepapers without an implementation should be considered paper 🙂

E: Satoshi had a proof sketch. And he turned out to be wrong. Ask me how I know. 🙂

A: doesn't really matter.  our goal is pbft, but evaluate N views concurrently, all message timeouts use a VDF based forced delay.  its not magic.  someone else can implement and prove it as well. the VDF removes any partitioning due to clock drift. N views allows you to amortize the cost of synchronization. @el33th4x0r can you imagine pbft that evalues 1 to N views at a time?  and one that uses a VDF for all timeouts (or a global clock), that's all we are doing. The latency speed up comes from the fact that most of the time, 2/3+ of the network seem to have packet loss below our erasure code ratio

E: Anatoly, PBFT is a specific protocol. It has an O(N^3) view change protocol. I don't think that's what you have implemented. I can take a look at your code at some point, but I think I know what I'll find. Here's a sanity check for everyone out there: For a deterministic asynchronous consensus protocol to be correct, there has to be a path in the code that can take forever. Literally forever. Practitioners the world over have invented protocols that are glorified replication protocols. They typically consist of a single loop or a doubly nested loop. I have worked with a well-known professor who did exactly this and his work is well known to be incorrect. It's an incredibly common mistake, one that smart people make.

A: the whole "trick" to performance is make the best case the average case.  almost all the time all the views are a direct descendant of each other.

E: If your protocol is deterministic and asynchronous, and there is no path in it that can take literally forever, then it is guaranteed to be incorrect. Period. This is literally the first lesson in graduate distributed systems. Is there a path in the Solana code that can take forever?

A: it halts if we have a 1/3+ partition for more than an epoch

E: That's different

A: which is 2 days

E: 👆

A: well, each node has a local random source

E: Irrelevant

A: assuming those stay truly random, and all data arrives eventually

E: 👆

A: if all data arrives eventually and every node has a secure random source?

E: Yes

A: i am pretty sure it will continue

E: That's not what I'm asking. This isn't a question about liveness. This is why this stuff is subtle. Let's try it again

A: is there a proof that we can't have pbft that evalues more than 1 view at a time? is there a proof that pbft can't use a global clock instead of local clocks? take those two things, you have our consensus implementaiton. if you can find a test that works on that system and fails on ours, we will fix the bug

E: The consensus code in Solana, does it look like linear, or like a loop, or a doubly nested loop? Is there a triple nested loop? Is there an open-ended loop that can take forever?

A: its more event driven, we evaluate the state once a new block is available. that computation is consistent for all the nodes. if they have the same data, arriving in any order

E: This is how my colleague got in trouble as well

A: you are shooting in the dark. make pbft handle more then 1 view a time and dont use weak assumptions about time

E: No, it's quite a pointed question. I already identified the core safety problem.

A: and you wont have a weak consensus algorithm 🙂

E: You keep using the word PBFT, I don't think it means what you think it means

A: emins proof of consensus involves how many loops you use 🙂

E: Correct. My proof of incorrect consensus very much involves whether the code has a potential infinite loop. This is a corollary of the FLP result

A: the linux scheduler?

E: That has nothing to do with anything

A: its an infinite loop

E: That's a common fallacy. Just because you have an infinite loop doesn't mean you have a correct consensus protocol. But if you don't have an infinite loop, then you cannot be a correct consensus protocol in a deterministic asynchronous setting.

A: linux scheduler is the infinite loop if you are data event driven

E: 👆

A: i think you mean, do i set local timeouts and take action on them in the future

E: No

A: well, thats equivalent to an infinite loop

E: That is a question about a synchrony assumption. I'm happy to have you make that assumption for liveness purposes. The actual PBFT paper is worth a read. It makes these issues somewhat clear.

A: nodes can't get into a state where they wont eventually produce a block. ie, propose a new view

E: Right. That indicates that your consensus protocol is broken.

A: that doesn't mean that the rest of the network will accept that view. thats true about every pbft implemetnation out there.  tendermint will alwasy try to produce a new block from the last known one. up until the epoch expires and the node halts

E: Tendermint has a proper view change algo of complexity O(N^3) and an infinite loop where it can get stuck forever.

A: yes, the linux scheduler will keep us stuck forever

E: It implements a well known algorithm known as PBFT.

A: waiting on new data

E: No, that's not what I'm referring to. I think I gave this a decent try, on a Saturday, spent with family.

A: 🙂

E: Perhaps someone else can help. I do believe that you have a fast something, but that something is certainly not BFT and in a BFT environment, will not guarantee safety.

A: fud 🙂

E: I do expect better from you.

A: is pbft that can evaluate N views at a time impossible?

E: The smiley isn't quite enough after I patiently went over basics. It's highly possible. Look into multi-paxos.

A: if you used some global time reference, would you not eliminate partitions due to clock drift?

E: There's a huge literature here that preferred Solana. And the last line of defense against errors is to produce a proof of correctness.

A: thats all we are doing.  if you can write a test that shows that are deviating from that, its a bug we can fix

E: This is why most people who have been in this area consider protocols without accompanying proofs broken. This is exactly the wrong attitude. The whole thing is a bug :-). Unless it has a proof. 🙂

A: thats FUD. so much fud

E: See I'm doing the smiley thing. 🙂 Seriously, every expert believes protocols without proofs to be broken. It's like rolling your own crypto.

A: it's all in good fun. i appreciate the feedback.  we have been super focused on implementation and less on correctness

E: "I have an encryption scheme, if you break it, it's a bug and I'll fix it"

A: i feel that so far we are good enough to be Consistent, and probably have a ton of halting bugs to fix. the design is still, pbft with N views, which is not a "fundemental" broken thing and use a global clock instead of local assumptions about time. our global clock is a VDF, tons of things to argue about there. Depends on the failure mode.  If the failure mode that the output is not decodable then it’s fine. Proving that the network is going to be consistent or halt within time T is easier

 

Ссылка на сообщение

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Гость
Ответить в тему...

×   Вставлено в виде отформатированного текста.   Вставить в виде обычного текста

  Разрешено не более 75 эмодзи.

×   Ваша ссылка была автоматически встроена.   Отобразить как ссылку

×   Ваш предыдущий контент был восстановлен.   Очистить редактор

×   Вы не можете вставить изображения напрямую. Загрузите или вставьте изображения по ссылке.

×
×
  • Создать...