 Show Posts
|
|
Pages: « 1 [2] 3 4 »
|
I support BIP100's voting mechanism above BIP101's one-way ticket to hell. Hard forks are still gonna have to happen in the future.
With BIP101 we will be at 32Mb limit not sooner than 2020, with BIP100's voting mechanism, however, we can get there already in a couple of years, but considering the hesitation of Chinese miners to go beyond 8Mb anytime soon and the psychological effect of there being a hard cap of 32Mb (if implemented properly), we likely (but not certainly) won't see any attempts to abuse and manipulate the block size beyond what the network can sustain with its current size. Hard forks is a good way to reconcile the current issues and exercise achieving consensus on how to resolve them. If done every once in a while, it's healthy for Bitcoin and crypto-currencies in general, as it puts most active members on alert and reminds us how to stay vigilant. The profits we derive from ecosystem gaining momentum must be well deserved. |
|
|
|
That is not what is meant by historical.
I mean, we don't know how historical limit on "network message" size (32MB) affects the size of a single block that can, for example, be relayed using multiple messages (now or in the future). We never had blocks near 32MB, so this part of the BIP100 is untested and not guaranteed to specify a consistent behavior across multiple implementations of Bitcoin protocol. It's either an explicit limit on block size (like 1Mb) or it isn't. To be clarified. |
|
|
|
If BIP100 is going to be amended, then fairly opaque statement The historical 32MB limit remains.
needs to be replaced with more explicit The original 1MB limit is replaced with the new 32MB limit.
to alleviate any confusion that may arise from what historical (never-approached, never-tested) limit actually means. It reminds me of the hidden configuration issue that caused the fork when switching from BerkleyDB to LevelDB some time in March 2013. Relying on some obscure historical legacy untested parameters is not a responsible decision, unless the intent is to slip a hole in there. |
|
|
|
we all just want agreement
Pretty much this. Hopefully there can be some kind of compromise reached in the next few months. If not hopefully one wins & the other dies so we can continue on the currently crumbling road that leads to the moon. The only middleground I can currently think of would be to gradually adjust max blocksize based on the avarage network load. Just like mining difficulty is readjusted every two weeks, the same could be done for max blocksize? It's a bit more complex than that. If there is an economic incentive for miners to agree on a bigger block size (to include more transactions with fees, for example), they will have little to no consideration for the ability of the non-mining full nodes to keep up with the load, as those are not explicitly incentivized to operate in the current setup. Having a static hard cap on block size (reflecting network's ability to sustain itself within its current size) is the only measure that non-mining full nodes have in order to counter potential misbehaviors by the miners. If the hard limit is chosen way higher than current technological capabilities and reached prematurely it may greatly reduce the amount of nodes participating in consensus and therefore destroy the decentralization premise that Bitcoin is standing on. |
|
|
|
I do have some concerns though.
The idea of sidechains has a lot in common with how first banks operated.
You park your gold (bitcoins) with a bank in a vault (sidechain) and you get your IOUs to participate in the economy.
At first, those IOUs were 1:1 redeemable back to gold, but that didn't last very long.
We all know, that history repeats itself, so this analogy is really hard to overlook.
The questions that arise here are these:
1) How hard would it be to break the original 1:1 peg or steal parked bitcoins while everyone is using IOUs?
2) How will the network handle the load in case of a bank run, when a lot of people suddenly want to unpark their bitcoins from the sidechain they no longer trust? What would be the cost to switch here?
3) How tempting would it be for governments to create national sidechains and law-enforce them as legal tender, now that mechanism for such enforcement is shaping up (while the load on the mainchain would be such that it will be prohibitive for small players to use, so it would be of no value)?
Good questions. I'll try to answer, but whoever knows sidechain technical details better (and epecially whoever knows the future  ) is welcome to correct me. 0) Sidecoins are not IOUs. They are bona fide coins, just like bitcoins. I would compare them with dollars of different denominations. Like 100$ bills vs quarters coins, they are different, but they are both dollars. Bills are more convenient for big purchases, coins are more convenient for micro-transactions , but both are basically the same and can always be converted to each other, at 1:1 rate. 1) If there is no bugs in the cryptocode, stealing parked bitcoins without parking corresponding sidecoins must be impossible. The same about breaking 1:1 ratio. (I use 1:1 for simplicity, the original ratio could be any. Like ratio of changing quarters for 100$ bills is not 1:1, but 1:400. What matters is that you can return it back at exactly the same ratio). 2) If I understand it correctly, no second party required for (un)parking. Converting bitcoins to sidecoins and back is basically sending them to some special addresses, that is, not much different from moving your bitcoin between two your addresses. How the network would handle it depends on network. If you are using the network for purchases several times a day then using it one more time to park all out should be no problem (the network doesn't care how much you are transferring, 0.1c or 1M$, the load is the same). However, it there is "main" blockchain that have to be used for unparking and if it that can handle only, say, one transaction per person per year, then congestion may happen. It would be good business opportunity for aggregation services although. Anyone can offer it: "send your sidecoin and I transfer it to any other sidecoin within an hour". Even at 7 transactions per second, there is 25,000 time slots in an hour, so 25,000 competing companies can offer such a service. With such level of competition, prices won't be too high. 3) The government is the permanent threat, with or without sidechains.  Yes, sidechain will make creating NSA-coin or IRS-coin much easier. But it will also make make escaping from them to some state-resistant coins, like Monero, much easier too. 4) "We" want different thins, as the current core dev split shows. I personally want and hope for bitcoin to kill and replace fiat. Where will be the mail volume of transaction, in bitcoin or on some sidechain, IMO, can't be told at this stage and doesn't matter much. Thanks for the answers. I'm interested to see how the hashrate would split between mainchain and various sidechains and what implications it would have on security of the whole system. I must admit, I haven't read sidechain's whitepaper yet, so maybe I will find my answer there. I would like to evaluate the proposal on its technical merit (when it comes out officially) in order to estimate how much of a drastic change that would be to Bitcoin and whether I would be willing to support it or not. However, if best Bitcoin developers are convinced to work on it full time, it might be worth a look. Even if Bitcoin decides to go another route, one of the alts might want to adopt it. Time will tell. |
|
|
|
-The historical 32MB limit remains. BIP 100 was better before Jeff re-added this limitation (it didn't exist in his original BIP100) and I'm concerned about the miner vote being easy to manipulate. On one hand, there must be a hard limit approximating network's technical ability to consistently converge on a single chain, on the other hand, it would seem appropriate for miners to decide on the current soft limit. Miners manipulating the soft limit should present no danger to the network's ability to operate. However, touching the hard limit without the overwhelming consensus of all full nodes must be prohibited. The hard limit also defines the potential size of the network, thus (mis-)placing it way above current technical capabilities of the average home connection will significantly reduce the amount of nodes participating in consensus (if limit is reached prematurely). I'm uncertain that 32Mb limit is safe enough to not create significant turbulence in the network's operation, if reached within the next 2-4 years. We are not necessarily talking about Chinese miners here as those are bandwidth limited already, but rather about potential rogue miners that might be interested in taking advantage of the situation. If you're concerned about miners' ability to manipulate the soft limit, you should advocate for lower static hard limit to counter that, but instead you seem to be disappointed by its re-introduction in BIP100, which sounds contradictory. Please clarify. |
|
|
|
|
In case BIP100 gains enough momentum and wins the race,
I would put its proclaimed static 32Mb limit in place of the current 1Mb cap explicitly (to be extra sure) instead of relying on some legacy network message limitations which may no longer fully apply to current (and future) block propagation mechanisms (blocks split across multiple messages, etc).
If that can be guaranteed to be implemented reliably, I would likely support BIP100 without much further hesitation, though I haven't studied it with enough scrutiny to know for certain that there aren't any holes there.
|
|
|
|
-snip-
Regarding, 32Mb limit. I'm not sure if it comes from the original network message limitation in Bitcoin or is a new addition related to BIP100. It was explained somewhere on this forum, that the block no longer has to fit in one network message (due to headers first implementation), so that original 32Mb limit may no longer apply. But better call the experts on this.
-snip-
It is not an BIP 100 addition. It was unlimited but p2p protocol limited maximum size to 32 MB. I would pay more attention to this before jumping to any conclusions too fast. If the code to implement BIP100 is complex enough there is a chance of a slip up somewhere and the whole thing might go belly up. The hard limit is important because it greatly reduces the incentive for rogue entities to start piling up the hashpower in order to manipulate the block size and push the majority of the full nodes out of capacity to operate and validate the blockchain. Just having a reliable hard limit has a psychological effect to not bother attacking, because the network cannot be destroyed through this mechanism alone (provided it is capable of operating at near full capacity). Whether 32Mb is a good enough limit to prevent abuse by (potential rogue) miners for the next few years is debatable (8Mb seems safer). We must make absolutely certain, that the 32Mb cap still holds (it may not be true in more recent implementations as blocks can occupy more than one network message of 32Mb) before expressing full support for BIP100. Edit: This would be a great solution. I also see that majority here likes BIP 100. Lest hope this will be pushed and that this never-ending debate will finally end.
According to some reddit sources, already 35% of all miners are in favour of this change. Lets hope!
 https://www.blocktrail.com/BTC/poolsThe 8Mb option is still holding up nicely. It might be more appealing to those who don't want to change Bitcoin too much. Bitcoin with high enough hard limit and gradually increasing soft limits is the one we're all used to and the one that is proven to work. The 8Mb option provides just that, only the numbers are different (hard cap: 8Mb, soft caps can be: 2Mb, 4Mb, 6Mb). Run this for the next four years, collect the data, evaluate the number of full nodes, reach consensus on how to move forward, celebrate the halving, be happy! That's the recipe for success right there.  |
|
|
|
What if China fixes their connectivity problems in the future?
We will then have the highest mining capacity and decent connectivity all in one place with a mechanism that allows miners to push the block size limit high enough to dominate the blockchain completely. Hard cap allows to observe network dynamic while staying on the safe side and then adjust the limit when the need arises.
First off, its the pools that vote, not the miners directly. Which, if you aren't aware of already, the majority is currently in China and they're fine with going to 8MB today even with the crappy Infrastructure. Second, there is a hard cap - 32MB - which has adjustable increments within it to handle active market pressure. Which is a very reasonable cap, getting us to roughly 20 million transactions per day. It will allow Bitcoin to surpass PayPal, and handle 1/8th the Visa transactions. A vote for XT would do more towards fueling your fears than this proposal does. Chinese pools are likely not running on a "home internet" connection, that's why they have evaluated that 8Mb is probably ok for them. But the ability to run a full node at home for the rest of the network would be a must if we want a healthy network of nodes. Fortunately, in Europe, US, and other developed countries home connections are capable of handling 8Mb blocks, though some users might suffer from this temporarily before the tech catches up. Regarding, 32Mb limit. I'm not sure if it comes from the original network message limitation in Bitcoin or is a new addition related to BIP100. It was explained somewhere on this forum, that the block no longer has to fit in one network message (due to headers first implementation), so that original 32Mb limit may no longer apply. But better call the experts on this. All in all, BIP100 seems fairly balanced to me, so I wouldn't be opposed to it, just thought that a 8Mb static cap was easier to implement without bugs and safer from perspective of manipulation attempts by the miners. |
|
|
|
|
Basically the main question boils down to this.
Do we want Bitcoin to be a single ledger that co-exists with current fiat and serves as a counter-balance to it?
Or do we want Bitcoin to eventually replace fiat entirely with sidechains?
My major concern is that once enough bitcoins are parked to sidechains, they will slowly but surely start to lose value, because no one is going to accept them directly and transact with them directly, as the time and cost of unparking them would make it prohibitive.
By the way, the single ledger idea doesn't necessitate that every coffee purchase is made on the blockchain.
The natural market for transactions shaped by the network's physical ability to process them will define what is in and what is out.
|
|
|
|
Maybe you guys "just don't get it" but I no longer run a "full node" because I live in China and the bandwidth doesn't work (so I would think that pretty much no-one in China runs a full node on a home-grade internet connection).
Although I am only using "home internet' I am sure that others in China would also have problems trying to run full nodes due to the controls of the internet here (so you are simply not going to see any Chinese nodes able to handle GB blocks with any reasonable speed).
Since a lot of hash-power is concentrated in China, it's appropriate that the majority of full nodes will then come from the rest of the world to help define what Bitcoin is. That's the balance I see. Chinese miners alone won't be able to replace the 6000 full nodes that exist in the network today, but they can collectively agree to implement a soft limit on block size in case a more aggressive approach gains momentum. From what I've heard, the move from 20Mb to 8Mb was pushed by Chinese miners due to poor connectivity in the region.
This is exactly what BIP100 gives them. By the mining pool voting with the blocks it produces, the majority can pick what the blocksize limit is. If the majority of Chinese pools don't want to go above 8MB, then they keep their voting at 8MB or under, and it's the top 80% of votes that judge the blocksize. BIP100 is the perfect compromise for the current situation. ... What if China fixes their connectivity problems in the future? We will then have the highest mining capacity and decent connectivity all in one place with a mechanism that allows miners to push the block size limit high enough to dominate the blockchain completely. Hard cap allows to observe network dynamics, while staying on the safe side and adjust the limit when the need arises, while at the same time exercising achieving consensus on what the next limit should be. |
|
|
|
Anything beyond 2020 is a best effort guess
True. In a system that values consensus, like Bitcoin, it would be natural for participants to exercise achieving it consistently on a regular basis. I suggest we re-evaluate the grounds that Bitcoin is standing on (with a possibility of a hard fork to rectify known issues) every 4 years. The "halving celebration" needs to be well deserved by having to achieve consensus on how Bitcoin's going to move forward. Raising the hard cap would be one of those issues on the table. In regards to this whole blocksize debate I think one further thing needs to be stated and that is "what are the Chinese miners going to do?".
The Chinese miners account for more than 50% of the hashpower so in effect it is really up to them what will happen (as much as that probably galls all westerners).
My guess is that they'll remain silent about the issue but will decide one way or the other and we'll find out about that early next year (so all the grandstanding going on now will actually amount to nothing).
(by "miners" I am meaning "mining pools" of course)
Chinese miners alone won't be able to replace the 6000 full nodes that exist in the network today, but they can collectively agree to implement a soft limit on block size in case a more aggressive approach gains momentum. From what I've heard, the move from 20Mb to 8Mb was pushed by Chinese miners due to poor connectivity in the region. |
|
|
|
Right now, miners cannot produce blocks larger than 1Mb (thanks to full nodes' consensus not to accept them), even if it is economically viable for miners to increase the limit, they can't. I suggest we jump from 1Mb to 8Mb hard limit and re-evaluate in about 4 years based on the data we collect.
Okay - yes I get what you mean now. I think that there might have been a BIP to do just that also. But understand that as less and less people run full nodes the miners will actually end up more in control (eventually I think all consensus will be decided by the miners only as there is no economic incentive for non-miners to run full nodes). Miners determine the longest chain, full nodes determine the valid chain. The longest valid blockchain is called Bitcoin, so both miners and full nodes have a say in what Bitcoin is. Statically capped Bitcoin with gradually increasing soft limits is Bitcoin we all recognize and value. Hope that stays. Miners also determine the valid chain because they also run full nodes. CIYAM stated it. -snip-
But understand that as less and less people run full nodes the miners will actually end up more in control (eventually I think all consensus will be decided by the miners only as there is no economic incentive for non-miners to run full nodes).
Miners normally constitute a small minority in the network of full nodes. It is the rest of the network (non-mining nodes) that prevents miners from forging the definition of "valid" chain. Thus letting miners alone decide on the block size limit gives them unprecedented ability to wipe out the rest of the network and redefine what "valid" chain is. |
|
|
|
Right now, miners cannot produce blocks larger than 1Mb (thanks to full nodes' consensus not to accept them), even if it is economically viable for miners to increase the limit, they can't. I suggest we jump from 1Mb to 8Mb hard limit and re-evaluate in about 4 years based on the data we collect.
Okay - yes I get what you mean now. I think that there might have been a BIP to do just that also. But understand that as less and less people run full nodes the miners will actually end up more in control (eventually I think all consensus will be decided by the miners only as there is no economic incentive for non-miners to run full nodes). Miners determine the longest chain, full nodes determine the valid chain. The longest valid chain is called Bitcoin, so both miners and full nodes have a say in what Bitcoin is. Statically capped Bitcoin with gradually increasing soft limits is Bitcoin we all recognize and value. Hope that stays. |
|
|
|
A static flat cap gives end users running full nodes a way to stay in the game and to limit the voting power given to the miners.
Huh - what on earth do you mean? You do understand that a "full node" that is not mining has "no stake in the game" don't you? You either keep running a full node or you don't (according to how much bandwidth you have) but no-one is going to care one way or the other (it is basically irrelevant whether or not you run a full node if you are not mining). Right now, miners cannot produce blocks larger than 1Mb (thanks to full nodes' consensus not to accept them). Even if it is economically viable for miners to increase the limit, they can't. I suggest we jump from 1Mb to 8Mb hard cap and re-evaluate in about 4 years based on the data we collect. |
|
|
|
...
I think that BIP 100 is a more intelligent approach as it allows the miners (the only people that really matter in regards to block sizes) to vote between themselves as to when an increase should occur (rather than just relying upon a theory that bandwidth will increase much like processing power).
...
I wouldn't give miners that much freedom with regards to the block size limit. A group of well-connected profit-driven miners would have little consideration for full nodes' capacity to handle the increased traffic. They (miners), in fact, might not even notice the drop in the number of full nodes, while playing with increased block sizes. A static flat cap gives end users running full nodes a way to stay in the game and to limit the voting power given to the miners. Static limit is really hard to abuse from that perspective. |
|
|
|
|
I do have some concerns though.
The idea of sidechains has a lot in common with how first banks operated.
You park your gold (bitcoins) with a bank in a vault (sidechain) and you get your IOUs to participate in the economy.
At first, those IOUs were 1:1 redeemable back to gold, but that didn't last very long.
We all know, that history repeats itself, so this analogy is really hard to overlook.
The questions that arise here are these:
1) How hard would it be to break the original 1:1 peg or steal parked bitcoins while everyone is using IOUs?
2) How will the network handle the load in case of a bank run, when a lot of people suddenly want to unpark their bitcoins from the sidechain they no longer trust? What would be the cost to switch here?
3) How tempting would it be for governments to create national sidechains and law-enforce them as legal tender, now that mechanism for such enforcement is shaping up (while the load on the mainchain would be such that it will be prohibitive for small players to use, so it would be of no value to them)?
|
|
|
|
Even under bitcoinXT's own parameters, only miners can vote for the fork to be triggered. Only one of the companies that signed this document is directly involved with mining. I've seen many people trying to argue that with those captains of industry showing support to BIP101 adoption of Gavin's and Mike's proposal is unavoidable. That's not the case however. Those businesses are not the entirety of bitcoin ecosystem, and even if we disregard the vote mechanism embedded in bitcoinXT miners are still a very important part of the bitcoin economy. Like it or not, Chinese miners are a huge part of the mining business and they have good reasons to veto bitcoinXT, BIP101 or even everything Mike Hearn is involved with. Given that Mike has said that bitcoin could do without China if the miners won't agree with him. If making the CEOs of those western companies sign this document in an attempt to make his worst case scenario sound more reasonable then all I have to say it that I'm disgusted. The good thing is that BIP101-enabled code (once deployed) would still follow the longest chain if the rest of the network decided to cap the block size at a static flat 8Mb. There simply won't be any block producers that would want to shoot higher than that. So there is no contradiction here. But, I do agree, that BIP101 seems to be designed to significantly shift the balance of power in the network from regions with higher hashing capacity towards the ones with higher bandwidth, which is much harder to increase quickly in order to counter some potential abuses that may arise from this shift. Plus, the effects of block sizes larger than 8Mb on network's structural integrity are unknown at this stage. We need to collect more data on network dynamics under the new hard limit in order to see how fast the technology is able to catch up and what other issues may arise from the increased capacity. For example, how the new limit affects the amount of full nodes and whether an average home internet connection is able to handle the load. Going with a static flat 8Mb limit might create some turbulence down the road, but eventually the tech will advance to the point that we will be "back to normal" with the new data to evaluate and decide how to move forward. |
|
|
|
The sky is falling! Bitcoin leaders and visionaries, friends, allies and collaborators have turned into a pack of fighting rats. Bitcoiners are watching helplessly how their former semi-gods are destroying their life-savings (Money is reputation and by killing reputation they kill bitcoin). And what is left will be taken over by NSA and replaced by banks' private chains. Why? Because of the consensus. Anyone who ever run a business or just has some life experience, knows: There Must Be Only One Boss. Many bosses - receipt for disaster. Because people tend to disagree. It's in our nature. What would happen if we demand consensus from core developers? It depend on how much do they care. If they don't care a lot, it'll be stalemate. I.e. nothing will be done. It was yesterday. If they do care a lot, it'll be a war. I.e. everything will be destroyed. It is today. So, one boss. But one boss means centralization. The boss will work not for us, but against us, for himself. Or for NSA. Problem we have. If we want anything done, there must be one boss. If we want anything done for us, everybody must be the bosses. How to resolve this dilemma? The solution is known for thousand of years: The Free Market. And it already exist in coinspase: altcoins. If you feel you know the right way, create your alt. Or join (and obey) any other creator. Or use anyone's alt. Creator is the boss because he is choosing what to create. And user is the boss because he is choosing what to use. The problem solved, hurrah! -Want big blocks? Here is an altcoin for it. -Want fast confirmations? Here is an altcoin for it. -Want anonymity? Want blacklisting? Want micro-transactions? Want whatever you want, there will be an altcoin for it, tailored for your needs and tastes. Except it doesn't work. We are not using this solution. Instead of sailing off happily on our favorite altcoin, we are all remaining hostages on this sinking bitcoin Titanic. Why? Because of cost of switching. If switching costs are low, customer is a king. He is in control. If switching costs are high, customer is a hostage. He is powerless. Switching costs in bitcoin are very high. Buying an altcoin is a bet. And considering the number of possible wrong ways (and the most of ways are wrong) - almost guaranteed losing bet. Switching is too high risk. And this risk turns us from kings to hostages. So, to make the free market to work, we have to eliminate the cost of switching. Fortunately, it is possible. What's more, it is almost done. I'm talking about sidechains, of course. If you bet on a wrong sidecoin, you won't lose a dime, because you can always switch back to main bitcoin blockchain or to any other sidecoin. At the rate 1:1. Which means that the risk (i.e. the cost) of switching is zero. As Antonopulos predicts, switching between sidecoins would be done directly in smartphone wallets. You may even don't bother to know, which exact sidecoins are used for this specific transaction. So all what have to be done is to: 1. Complete the sidechain BIP (along with bigger blocksize, to cut paranoia level ) 2. Freeze the bitcoin core code. The frozen core will give stability and sidechains will give flexibility. And let a 100 flowers sidechains blossom! While the miracle of free market will turn fighting rats and evil gods back into our best friends. tl;dr: Sidechains (i.e. blockstream) will enable free market in bitcoin development. It'll save Bitcoin.Some good points here. Would be interesting to see how it develops in the future. |
|
|
|
Most of them support either 8MB or Garzik's BIP100. Gavin's and Hearn's BIP101 is getting extremely little support (less than 1%). This is because even the most business-minded CEOs of funded startups who dream of taking on board 2 billion users in 6 months understand that going to 8MB to then double the block size every fucking year till we reach 8GB blocks (!!!) is simply reckless and crazy. Almost everybody knows that bigger blocks are needed rather sooner than later - what the SANE people in this community is trying to do is to reach a reasonable and safe consensus about how and when to increase such limits. Finally, I guess that even the CEOs of hyper-funded startups understand what bitcoin users are looking for when they use bitcoin... Quoting a thread on reddit: I want an incorruptible non-government controlled store of value upon which a new global currency can be built. I don't care about low transaction fees or fast confirmations as I already have those. I want a place to store my wealth that can't be stolen via inflation of the money supply. A wealth asset that is safe and can be converted to any medium of exchange that I want when I choose. I want a digital gold. I want privacy. I want a money which is not centrally controlled but rather controls the growth of centralized institutions and gives more power back to individuals enabling us to make economic judgments and allocate scarce resources in a distributed manner not in a centralized one. I think this is what Satoshi also really wanted and I think we're gonna get it. Yep, that's what concerns me about BIP101: we are going to be at 12Mb limit in the beginning of 2017 and at 16Mb limit in the beginning of 2018. It increases gradually over time and seems overly optimistic, and it's not clear if the network can sustain itself without any limit at all. My guess, is that having a static hard cap has more of a psychological effect on preventing attacks than anything else, because an attack cannot have long lasting effects with a known ceiling on block size, so attackers simply won't bother. The 8Mb cap, on the other hand, looks like a reasonable compromise, at least from the standpoint of being competitive as a transactional medium. With 8Mb blocks Bitcoin will have only twice the pre-fork capacity of its closest PoW competitor, so going for anything less might endanger Bitcoin's position as a leader in this space. |
|
|
|
|
