CPS-0004? | Spending Script Redundant Execution

[fallen-icarus]

(rendered proposal in branch)

[rphair]

@fallen-icarus we need to reduce the amount of non-essential words in the title while still covering exactly what this is about. Would it be accurate & complete to title this Spending Script Redundant Execution? If not, then something else with a small number of meaningful words?

[fallen-icarus]

@rphair Spending Script Redundant Execution works for me. I'll change the title.

[L-as]

I'm not sure this is an actual problem? If you have this problem, your script should rather than doing the computation, check that you withdraw 0 ADA as rewards from a helper script that checks the computation.

[fallen-icarus]

@L-as I'm not sure what you mean. This has nothing to do with rewards. Can you please elaborate on what you mean?

[L-as]

@L-as I'm not sure what you mean. This has nothing to do with rewards. Can you please elaborate on what you mean?

It's a trick to solve exactly the issue you describe. You can always withdraw 0 ADA.

[fallen-icarus]

It's a trick to solve exactly the issue you describe. You can always withdraw 0 ADA.

My understanding of this trick is the following:

Script A	Script B
checks whole transaction	only checks if Script A was executed
has ADA that is never actually withdrawn	has any number of utxos
executed once per transaction	executed once per utxo from Script B

If this is correct then this trick does not solve the issue. Script B will still redundantly check if Script A was executed. This does not stop redundant script executions; it only minimizes the cost of those redundant executions.

Also this trick only works well with centralized smart contracts. I have created a DEX proof-of-concept (that I will open source soon) where users maintain full delegation control of their assets at all times. As a design requirement for this feature, all users have their own swap contracts. Users can then lookup all available swap contracts to remotely interact with other users' swaps. I have already extensively tested the DEX and it works as expected. It also has other nice properties such as being naturally concurrent (there are at least as many utxos as there are users) and therefore doesn't require batchers. The only issue is having to tolerate the redundant executions.

If I were to use this trick with the DEX, it would literally double the number of smart contracts required. Not only would this dramatically over-complicate the DEX design, but it would also create bottlenecks in the lookup process of finding all the necessary information for users to interact with the DEX. It is also worth pointing out that each user needs to store their reference scripts in each of their swap contract address. Right now this is like a deposit of 24 ADA in order to open a swap position. Using this trick would increase the necessary deposit each user needs to make per swap position.

The above drawbacks aren't an issue for centralized smart contracts. But these centralized smart contracts also have the drawbacks of:

1. Users must give up delegation control.
2. Batchers (middle-men) are required to overcome the lack of concurrency.

It is my strong opinion that, given the redundant executions incentivize this trick (in the pursuit of lower fees) and this trick strongly favors centralized smart contract designs, this actually further highlights why these redundant executions are a problem.

[rphair]

@fallen-icarus I really hope you will include this last explanation in your CPS somehow. Cardano has a great potential (compared to other L1's) to support fully distributed architectures and the DEX example is an excellent way to bring this problem statement into active discussion & solution.

[fallen-icarus]

@rphair I will look to edit the pull request this weekend and incorporate it. I was not aware of this trick before so thank you @L-as for mentioning it.

[L-as]

I'm not sure I understand your issue, but you definitely shouldn't have a script per user. I'm not sure how you came to that conclusion. I'm 'aso not sure how this pertains to batchers, or delegating, or centralisation.

In any case, I don't think the fee overhead from this trick is worth messing so heavily with how validators work.

[fallen-icarus]

@L-as Imagine you have the following setup:

Script A	Script B
checks whole transaction	checks if script A is executed
has 5 utxos at the address	the number of utxos is user dependent
executed once per utxo from this address	executed once per utxo from this address
script is used as a reference script	script is used as a reference script

Now imagine that Alice and Bob both have 5 utxos locked by Script B (10 utxos in total). What would happen in the following contexts?

1. Alice and Bob want to send all their utxos together in the same transaction
2. Alice and Bob want to send each of their utxos individually (10 txs in total)

In the first context, there is no major problem:

1. One utxo is used from Script A to force its execution which costs 0.3 ADA.
2. Script B redundantly checks if Script A is executed once for every utxo (10 times) but the cost of each check is minor at only 0.1 ADA.
3. The total fee is 0.3 ADA + 10 * 0.1 ADA = 1.3 ADA.

In the second context, there is now a bottleneck problem: each of the 10 txs need to use one utxo from Script A to force its execution but there are only 5 utxos available. In this context, batchers are required to use the limited number of Script A utxos efficiently.

You can increase the number of utxos at Script A but the concurrency will always be limited to the number of utxos at Script A. Given the world that blockchain hopes to achieve where everyone is using DeFi, would 1000 utxos be enough? What about 10000? If you guess wrong, bottlenecks will cause issues that only batchers can solve.

The reason for this bottleneck is due to Script A not being a real reference script. It is basically a frankein-script: half-reference, half-real. By this I mean that a completely unrelated (to the transaction) utxo needs to be consumed in order to "reference" Script A.

---

Now imagine this counter setup:

Script Alice	Script Bob
checks whole transaction	checks whole transaction
the number of utxos is up to Alice	the number of utxos is up to Bob
executed once per utxo from this address	executed once per utxo from this address
script is used as reference script	script is used as reference script

Imagine Alice has 5 utxos locked by Script Alice and Bob has 5 utxos locked by Script Bob. What would happen in the same two contexts before?

In the first context, there are annoying redundant checks that cost about 0.3 ADA each but it works:

1. 5 utxos reference Script Alice and 5 utxos reference Script Bob
2. The total fee after the redundant checks is 3 ADA. If there were no redundant checks, the fee would be 0.6 ADA (0.3 for Script Alice + 0.3 for Script Bob)

In the second context, there are no problems at all (there are no redundant checks since they are separate txs):

1. 5 transactions reference Script Alice for a total fee of 1.5 ADA.
2. 5 transactions reference Script Bob for a total fee of 1.5 ADA.
3. The total amount of fees is 3 ADA.

These 10 txs can go through at the same time because their is no concurrency bottleneck. Script Alice and Script Bob are always available since they are being used as true reference scripts.

---

The second setup's maximum concurrency is directly tied to the the number of utxos each user wants to use. In other words, the more people want to use it, the more concurrency is naturally available.

The first setup (which is the trick) has the concurrency "hardcoded" by the number of utxos at Script A. This concurrency does not naturally grow or shrink with the demand. In low demand contexts, there are likely to be too many utxos at Script A which results in inefficient use of the ADA locked at Script A. While in high demand contexts, there are likely to be too few utxos at Script A which means batchers will need to be used to compensate. The backlog that SundaeSwap suffered is a perfect example. Backlogs only exist when the concurrency is too low for the demand. This doesn't even deal who decides how many utxos are available from Script A which is another whole "can of worms".

In short, this trick sacrifices concurrency in order to minimize the cost of redundant executions. To compensate for this drop in concurrency, centralizing solutions are used, such as batchers. That is how this pertains to concurrency and centralization.

---

Edited to include fees for the first setup too.

[fallen-icarus]

@L-as

you definitely shouldn't have a script per user.

I have officially open sourced my proof-of-concept which can be found here. Given your stance, I would like to know your thoughts on it.

[L-as]

@fallen-icarus

1. Liquidity pools can be designed such that they are inherently decentralised, concurrent, and stakeable (to a liquidity provider chosen SPO).
2. Your design is sensible, but
   a) I don't see why you have a script per user. This also messes with reference scripts. Just put the information in a datum.
   b) Your design seems not to need to be on-chain, because there is no global state. You could do the swaps using state channels instead, or at the very least just do normal atomic swaps. There is the question of how to broadcast available swaps, for this you might want a sidechain (easy but nonoptimal), but you could also have a simple system where a P2P network disperses trades by each node broadcasting an open trade to its peers. You can make it such that it's profitable to pass on an open trade by cryptographically securing a fee if the trade is closed (through you). This is much more useful and practical on Cardano than e.g. Ethereum because assets are native and we have UTXOs.

[L-as]

Anyway, I don't get how this pertains to the proposal, and would say that maybe what you want is to optimise the script for checking the 0 ADA withdrawal. AFAICT though this can be a script of around ~100 bytes if not smaller. #309 is also relevant.

[fallen-icarus]

@L-as

I don't get how this pertains to the proposal

From my perspective, your view that redundant executions aren't a real issue stems from your set beliefs about how Dapps can and cannot be created. I thought if I showed you concrete proof of another (and in my option better) way, I could change your mind about this. I actually built the DEX so people can try it out. It has all the following desirable features:

1. Full user delegation control
2. No impermanent loss
3. Naturally concurrent
4. Easily upgradable
5. Easily integratable into all wallet software

Again in my opinion, these features make this proof-of-concept probably the best DEX that so far exists in DeFi. The only thing holding it back is the redundant executions. Without the redundant executions, it would also be much cheaper than any DEX using the "withdraw 0 ADA" trick.

As for your points:

Point 2 is addressed in its entirety in the README. Both the reference scripts and the available swaps are easily broadcasted by what I call Beacon Tokens. There is no need for any of the complicated techniques that were mentioned. It easily works on Cardano as is. As a bonus, these Beacon Tokens can be generalized to any Dapp. This means broadcasting any information to other users is trivial for any application. Also the need for a global state is not the only reason to use a blockchain. The blockchain is also meant for any peer to peer application where users do not trust each other. This latter reason is why my DEX is on-chain. There is no need for state-channels; there is plenty of room on layer 1.

With the current design of the DEX, there is basically a deposit required for opening swaps which comes from having the user store his/her reference script on-chain to then be found with the Beacon Tokens. This deposit also disincentivizes opening empty swap addresses. However, if the deposit is deemed to be too high (currently about 20 ADA), I also mention a potential adaptation of the design at the end of the README that does not have this high of a deposit requirement. I am mentioning this to point out that, combined with the Beacon Tokens, there are no drawbacks from giving users their own scripts/addresses.

As for point 1, I have looked at every DEX I could find that uses liquidity pools and none of them meet the gold standard of users maintaining full delegation control. By this I mean that, while both Alice and Bob use the DEX:

1. Alice can delegate her assets to ANY pool she wants.
2. Bob is not forced to delegate to the same pool as Alice. He too can delegate to ANY pool.
3. No one is forced to pick from a chosen subset of pools.

Liquidity pools can be designed such that they are inherently decentralised, concurrent, and stakeable (to a liquidity provider chosen SPO).

This still sounds like it violates the third requirement of the standard. From a first principle perspective, I am very skeptical of the idea that liquidity pools can ever be designed in a way to match the level of decentralization obtained from full user delegation control. Unless you can directly point to an actual liquidity pool based DEX that meets the above standard (including the third requirement), we will just have to agree to disagree.

In short, it seems to me that there are three consequences (that I know of so far) to having these redundant executions:

1. Wasted computational resources that result in higher fees.
2. It incentivizes a "trick" that creates concurrency bottlenecks which requires batchers to handle.
3. It holds back, in my strong opinion, better designed Dapps that can't compete with the fees of Dapps using the "trick" in consequence 2.

You keep suggesting to use the "withdraw 0 ADA" trick and I keep explaining why I do not believe this trick is a good solution: it leads to more centralized designs. I strongly disagree with the view "we should just use the trick instead of fixing the redundant executions". Also the problem is not that the execution costs are too high; the problem is that there are redundant executions being done so the fee compounds. 0.3 ADA is already extremely cheap to check a whole transaction. If I wanted to, converting my script to be used like the trick suggested is not an issue.

Therefore, #309 would not be useful. However, I think my proposal would actually solve #309 issue. My proposed solution allows a plutus script to be able to handle both the case where a datum is missing and the case where a datum is present; no proxy-scripts are necessary. It does this without having to create an on-chain encoding for a missing datum. I know there are other proposals motivated by the no datum issue. I will try to aggregate them and add to my proposal how they are easily solved by my proposed solution.

I do not mean to come across as shilling the DEX proof-of-concept. There is actually no way for me to personally profit from others using the DEX, aside from new people coming to Cardano to use it. (Cheers to being properly incentivized lol). It is for this reason that I think it is okay for me to discuss it. I personally believe it adds weight to the need for this proposal but if others feel the DEX does not help explain why the fix is needed, I can drop it from the discussion.

[rphair]

if others feel the DEX does not help explain why the fix is needed, I can drop it from the discussion.

In my opinion, the divergence of opinion is exactly why it needs to remain in the discussion and certainly in the CPS itself (I said that above, but would say so even more emphatically now). @fallen-icarus - others are going to want to follow your more "granular" (I don't know if that's the right word?) approach even if trade practice suggests otherwise.

Even if it were only an academic consideration it would still have a place in this CPS: editorially I can say that our overall process is about academic concepts finding implementations: sometimes gradually.

[L-as]

I have a lot of thoughts, but I will focus on the biggest issue:

• I don't see how this trick pertains to batching at all. Using this trick has nothing to do with batching, centralisation, or protocol design. Given a script s, you can turn it into s' that checks that 0 ADA is withdrawn from s'' with the same logic as s. This is a transformation that is valid for any script.

[fallen-icarus]

@L-as You are not accounting for the utxo ID changing every time the "withdraw 0 ADA" script is executed. On Cardano, a script will NOT execute unless a utxo from that script's address is consumed. It doesn't matter if there was no value taken. The old utxo will be destroyed and a new utxo with a different utxo ID will be produced. This will happen even if 0 ADA is actually taken from that script's address. It is NOT possible to successfully execute a script WITHOUT changing the utxo ID of the utxo that was used to force its execution. In order to force its execution again, the new utxo ID is now needed.

This is completely different to the way actual reference scripts work. Actual reference scripts assume all utxos actually being used are at the address that belongs to the reference script. To have two different scripts means you have two different addresses. In order to execute both scripts in each transaction, at least one utxo must be consumed from each address. This means the utxos IDs at the "withdraw 0 ADA" script will be constantly changing.

Are you thinking of how it works on another blockchain and thinking it will also work on Cardano?

[L-as]

@fallen-icarus OK, I think I understand: You mistakenly believe that you need a UTXO to use the withdraw 0 trick. This is not the case. There is no UTXO at all locked by the script (or using the script as staking address, but even if there were, it doesn't matter).
The trick simply involves checking that a stake script is withdrawn from. The amount doesn't matter (but it's important to note that it can be 0).

[fallen-icarus]

@L-as What is the cardano-cli transaction build option for this? Is it --auxiliary-script-file?

Edit:
I just tested it and I believe this is the command you are referring to. I missed it before. With this you might be right about not needing batchers but I am going to test it out. It might still create other problems for the DEX since giving users their own personalized spending scripts does more than just allow for full delegation control. I guess this CPS is on hold until I finish trying it out.

Thanks for pointing out my mistaken assumption.

[fallen-icarus]

So I have been testing using the --auxiliary-script-file option with the dex but I can't seem to find a way to get the auxiliary script to be seen by the plutus script. It doesn't seem to be in the ScriptContext that plutus scripts see.

I don't believe I mentioned this before but the dex uses composable atomic swaps. Do you want to convert 10 ADA into 5 WMT and 5 DUST? No problem; you can do this in one tx. Do you want to convert 5 ADA and 5 WMT to 3 AGIX and 7 DUST? Again no problem. This can also be done in the same tx.

Using composable atomic swaps means, in one tx, you can easily do:

many assets -> many assets

In order to maintain composability, there is a separate auxiliary script for each possible trading pair (easily created on the fly by users). Without the atomic swap contracts being able to verify that the required auxiliary script was executed, the security of the composition is completely broken.

@L-as Am I missing something else that would allow me to do this?

[fallen-icarus]

@L-as After thinking about it more, I'm starting to think the auxiliary script is not what you mean.

your script should rather than doing the computation, check that you withdraw 0 ADA as rewards from a helper script that checks the computation.

The trick simply involves checking that a stake script is withdrawn from. The amount doesn't matter (but it's important to note that it can be 0).

Given a script s, you can turn it into s' that checks that 0 ADA is withdrawn from s'' with the same logic as s.

Do you mean:

1. The "helper script" is literally a staking script, not a spending script.
2. The logic I need executed is done by the staking script.
3. The spending script then checks the txInfoWdrl part of the ScriptContext to see if the staking script was executed and 0 ADA was withdrawn from the reward address.

Is this correct?

[L-as]

That's what I've been saying from the start. I have not misspoken.

[fallen-icarus]

@L-as

I have not misspoken.

I know. The misunderstanding was my fault. I didn't realize you could literally use a staking script as a spending script like this. So whenever you mentioned rewards or staking, I thought you just didn't understand my issue. Apologies for this misunderstanding.

I will test this today and comment with what I find.

[fallen-icarus]

Okay so I was able to get everything working using the trick. However, despite the dex technically still working, I believe this CPS should still be pursued for the following reasons:

1. This trick means everything requires two scripts now instead of just one. While the spending script can be made quite small, this is still a drawback IMO.
2. This trick does not stop the redundant executions since the spending script will still redundantly check if the required staking script was executed. This trick just minimizes the cost of those redundant executions.
3. This is not how staking scripts were intended to be used.
4. My proposed solution to the redundant executions also fixes the datum issues other developers have been struggling with: it would make it possible to spend a utxo at a script address despite the utxo missing a datum or having the wrong datum. I believe the solution does this without requiring a change to the way things are stored on-chain. The only thing that would change would be the way plutus scripts are executed.

I don't think the ... overhead from this trick is worth messing so heavily with how validators work.

@L-as Maybe reasons 1, 2, and 3 alone are not enough of a reason to change how validators work. But when you also consider the possibility of reason 4, I believe it is worth doing. Do you?

[fallen-icarus]

@imikushin

Cardano documentation is quite... decentralized.

🤣

I'd perhaps want to use TxInInfo instead of TxOutRef

Something like this was suggested in another PR discussion.

Also, I don't think we need the datum resolution in TxOut since it is already in the txInfoData map.

[michaelpj]

I'm sorry, I'm a bit lost. Are you proposing changes to the ScriptPurpose independently of the grouped inputs proposal? The grouped inputs proposal is definitely not simple, and needs a lot more details worked out (and TBH I doubt would be accepted without a lot more motivation).

[imikushin]

@michaelpj

I'm sorry, I'm a bit lost. Are you proposing changes to the ScriptPurpose independently of the grouped inputs proposal?

In my comment #418 (comment) I was suggesting to @fallen-icarus that TxInInfo would be nice to use instead of TxOutRef in his grouped inputs proposal. Sorry if it's confusing.

[rphair]

@fallen-icarus (on Dec 31, 2022): I have officially open sourced my proof-of-concept which can be found here. Given your stance, I would like to know your thoughts on it.

[rphair]

Would it be fair to ask for a more detailed explanation of the P2P Atomic Swaps use case? This may be naïve (and I'm not technically capable of joining the debate of the proposal particulars) but if a treatment of this issue is helpful in implementing atomic swaps on Cardano, then I believe this use case should be covered in detail in the document... to progress this CPS and any CIP that facilitates that treatment.

I'm trying to get more reviewers based on interest in the Atomic Swap idea as you've described it on GitHub. If in agreement with what I've said above, I also think this repository should be linked from the CPS itself. Here is what I posted about fallen-icarus/cardano-swaps (please correct me if I've described the issue inaccurately or naïvely) in the Cardano MBO on Matrix:

I've been watching this concept develop from #418 and I don't think third party reviewers have been giving it enough credit, or even enough attention. I don't know every technical underpinning of every current & proposed DEX on Cardano but in the meantime this idea seems unique to me. I do wish I could do more to promote this idea since my own technical shortcomings make it impossible for me to convincingly push ahead any proposed standards that would be required to make this happen. I can only hope key people at IOG, CF, and Cardano dev agencies can take a serious look at how significant it would be for the industry & Cardano's own marketplace to support a true non-custodial DEX. 😯

https://matrix.to/#/!QvtUklhUfmEuCBIexz:matrix.org/$lU0OwYOAQXpNYVFJl1I33YPhhQ_wH14Jzu5etQP9Drk?via=matrix.adamantium.online&via=matrix.org&via=forum.balanceanalytics.io

[fallen-icarus]

Would it be fair to ask for a more detailed explanation of the P2P Atomic Swaps use case?

I fully intend to do this. However, I believe these redundant executions impact all distributed dapp designs, not just atomic swaps. In addition to this, I think people are underestimating the usefulness of atomic swaps; I believe you can slightly alter the logic from cardano-swaps to create a distributed p2p lending protocol. To prove it, I am currently working on a p2p lending proof-of-concept that does this. I have already finished the on-chain code for it; I just need to finish the helper cli program and the documentation. I hope to be done with it within a week or two. After that, I have an idea of how to create a distributed DAO where users maintain custody and delegation control while still enabling the DAO to spend funds from user wallets when certain conditions are met (such as a multisig). Again, it uses a similar design to cardano-swaps. The DAO design is pretty simple and I should hopefully finish it within a week or two of starting it.

All of the proof-of-concepts are impacted by the redundant executions. I would like to finish both the p2p lending and the distributed DAO implementations so that I can really emphasize why this problem should be fixed. I intend to link all three proof-of-concepts in the CPS once done. I appreciate your helping to spread the word and I agree that (until zhekson1's recent tweet) the design was not being taken seriously enough, although, I understand why since this idea is coming from an unknown community member. If there is enough of a demand, I can look to rewrite the CPS before finishing the other proof-of-concepts. I could possibly still explain the p2p lending design and the DAO design, just without being able to point to working prototypes.

[imikushin]

Just to add a little more color to this discussion: transaction-level scripts would trivially address problems like double satisfaction — just by being able to look at all sides of the trade at once.

I do realize that, of course, a per-output script execution also has access to all spent outputs, but: if, in general, we may have to look at all participating outputs anyway, why do per-output calls at all?

[imikushin]

I would perhaps even go as far as to suggest that any script doesn't need to be run more than once per transaction.

This means, that even if a script could be used for multiple purposes within a single transaction (e.g. mint and spend), it should only be called once.

This would improve both security and scalability.

[L-as]

Just to add a little more color to this discussion: transaction-level scripts would trivially address problems like double satisfaction — just by being able to look at all sides of the trade at once.

Trivially not true given that two different scripts can also cause the problem.

[fallen-icarus]

two different scripts can also cause the problem.

I don't see how this invalidates what @imikushin said. Since the script can see all the other inputs, you can either 1) disallow transactions that have inputs from other scripts or 2) keep track of the value from each input and track where the value goes.

[imikushin]

Trivially not true given that two different scripts can also cause the problem.

@L-as Well, now we have this problem even for one script. What I'm suggesting would eliminate it (for outputs spent from the same script), wouldn't it?

It would also make multi-script transaction logic (if you're into that) a lot simpler.

[SebastienGllmt]

Although this CIP is for a different use-case than CIP38 there is some overlap and I just added this to the CIP38 motivation section #309

[rphair]

@KtorZ @fallen-icarus @zhekson1 I'm trying to follow up on increased interest on the Cardano MBO on Matrix in Cardano Swaps and whatever it would take to implement them. I've tried to stay ahead of this idea and see it's fallen quiet for about a month along with some of the CIP documented components that would be required.

Can we summarise here, so the community can follow potential implementations and so we can keep moving this CPS forward:

1 - Is there a "best" place on the CIP repository for discussing Cardano Swaps? Or given that it's given a full heading in #466, would that be a better choice?

• On that note, if Cardano Swaps is a major use case for this CPS, shouldn't it have its own section rather than a single line in a bulleted list of Use Cases it has now?

2 - Should Cardano Swaps have its own CPS? ... e.g. if it's perceived that the reference implementation on GitHub might be done better with support in Plutus for an improved implementation? (cc @michaelpj)

3 - Whether we can designate such a place, or need to create a new one (like a GitHub Discussion, if that's not a step backward), can we then use that place to gather public discussions like the ones I started collecting in #466 (comment) when they relate to community interest & potential implementations.. both for the Cardano Swaps and related use cases?

4 - The SPO Digest today mentioned Cardano Working Groups: how to get involved which lists an IOG group "Developer experience" which would cover applications for DEX and this CPS's other use cases... but without any more specific (sub-?) groups for these areas.

• @KtorZ how would we go about suggesting that a working group be formed to address this similar set of use cases, starting with the arguably most urgent case of Cardano Swaps?
• If IOG hasn't formed or can't form a group or sub-group for it, is this something the CF could do?

Whether there are Plutus changes that need to happen first, or this CPS needs to be changed in scope, or an additional CPS needs to be created for Cardano Swaps or other DeFi components, etc... then such a Working Group might finally consolidate all this potential into better documentation of who needs to do what & in what order.

[fallen-icarus]

I do not use matrix so I cannot see the activity there. Here are my thoughts though:

1. zhekson1 and I are currently working on cardano-loans. As the name suggests, it is a lending and borrowing platform using a similar distributed dApp design as cardano-swaps. It has gone through several iterations but is in the final steps. I am cleaning up the code and zheckson1 is working on the readme. We are doing this in a private repo because we thought that would be cleaner since it has gone through so many iterations; we didn't want people thinking it was finalized before it actually was. That is why the last month "appears" quiet. cardano-loans is relevant to this CPS but I think it is better to explain how after you see how it works.
2. Both cardano-swaps and cardano-loans are written using IOG's PlutusTx. While great for testing and prototyping, it is not very efficient. @OpShin did some benchmarks with the various smart contract languages and found PlutusTx can be 100x less efficient that something like aiken. With that in mind, I think both cardano-swaps and cardano-loans should be rewritten using aiken to determine just how efficient this design currently is, despite the redundant executions. This is totally arbitrary, but IMO cardano-swaps probably only needs to compose 10-20 swaps in a single tx. There are diminishing returns for more than that. Given the aforementioned benchmarks, I think this is already achievable if it is written in aiken. I fully intend to rewrite the code using aiken once cardano-loans and one other PoC are finished. That being said, cardano-swaps was open sourced and explained so that someone else can take a stab at it. The same is true for cardano-loans. I don't think this CPS can be completed without first seeing how far another smart contract language can push the designs while still suffering from the redundant executions. Unfortunately, these PoCs are taking longer than I expected. I am already working on them full-time.
3. As for where to discuss these designs, I can look to add a discussion option to each PoC repo. I've never done something like that before but I'm willing to see how it goes. Perhaps the Issue section can be used like this, too? I am open to other ideas.
4. My thoughts regarding the PoCs and this CPS keep changing as I write them. This is another reason why this CPS is basically on hold until after the PoCs are done.

[rphair]

sounds great: not marking Waiting for Author yet, since it seems to me some principles of this CPS can be discussed which are separate to the PoC's that you are working on. 😎

[fallen-icarus]

After writing both cardano-swaps and cardano-loans in aiken, I am no longer concerned about the redundant executions. cardano-swaps can handle 8+ swaps in a single transaction, and cardano-loans can accept 6 loans or make payments on 8+ loans in a single transaction. I think these benchmarks are more than sufficient for right now. To be clear, redundant executions are still present but aiken is efficient enough to make them negligible. Therefore, I do not see a need to pursue this CPS.

[rphair]

That's very good to hear @fallen-icarus and so closing this would be consistent with others' suggestions that more general methods would be applicable, and thanks for the update.

Is there anything that would be helpful in standards, documentation, developer support, or visibility to Cardano official agencies that would help with cardano-swaps as I was hoping to find in #418 (review)?

If so, l'd be happy if you would open a CIPs issue to help get whatever support you might need for the Cardano Swaps, Loans, and other use cases you've documented here.

[matiwinnetou]

@fallen-icarus I would disagree with this actually. I have been playing with Hydra and L2 and some more complex use cases we totally blew up Aiken and Plutus.

This still makes sense and it all depends on complexity of what one is doing.

[fallen-icarus]

Is there anything that would be helpful in standards, documentation, developer support, or visibility to Cardano official agencies that would help with cardano-swaps as I was hoping to find in #418 (review)?

@rphair I would like to finish the feature set of these protocols before personally revisiting CIPs. The ideascale proposal pretty much covers what I would like to do. I never thought these POCs would take so long... There are a few issues where CIPs could help but I think the arguments for them would be stronger once there are fully featured protocols to point to. The issues mostly revolve around maintenance of these protocols. Will the official location for the code be my repo, the CIPs repo, or an entirely new repo like a "Cardano Protocols" repo? Should an entity like the cardano foundation hold the license to them? What will be the process for making changes to the protocols? I think these questions relate to the ones you asked in the linked comment. Perhaps the community can start thinking about these questions while the protocols are finalized.

[fallen-icarus]

@matiwinnetou Considering I have not personally dealt with the issues you mention, I do not think I am the right person to head this CPS if it should still be pursued. Should this one be re-opened or should a new PR be made where someone else is the author?

[rphair]

@fallen-icarus:
Will the official location for the code be my repo, the CIPs repo, or an entirely new repo like a "Cardano Protocols" repo?

So far we've determined that CIP related code (practical & reference implementations; examples are OK) should not be in the CIPs repo for a few reasons highlighted in other PRs:

• code changes (even for instructive examples) require PRs that have to funnel through editors who might not be the best to approve or improve them.
• it's easier to subscribe, star, file issues, etc. for a project in its own repository that blur this development with the CIP repo divided among a lot of different documentation efforts.
• the CIP process is focused on documentation rather than development, in addition to other subjective reasons.

Should an entity like the cardano foundation hold the license to them?

I can think of no practical goal that could be achieved any more easily by making that assignment. Anything the community does to promote Cardano P2P markets can rely entirely upon open source licenses published by developers themselves... and anything the CF does to promote these would not depend upon its holding license over either the code or its documentation in any CIPs.

@matiwinnetou Considering I have not personally dealt with the issues you mention, I do not think I am the right person to head this CPS if it should still be pursued. Should this one be re-opened or should a new PR be made where someone else is the author?

What do you say Mateusz? Upon your request we can open this up again and declare you the advocate of this CPS, with you eventfully adding yourself a CPS author whenever that might be appropriate. 🤔

[matiwinnetou]

@rphair - I am happy to take over this CIP as an author. I personally don't think having Aiken (which is very performant indeed) is an excuse not to purse this work in future versions of Plutus.