Re: Tau-Chain and Agoras Official Thread: Generalized P2P Network
February Tau monthly update video is now out!
https://youtu.be/oYRmJqhLp5g
Includes QandA as usual (QandA text summary can be found on our subreddit)
Tomas
- Adding features which enables a rewrite of the Prototype bot
- Initial networking features added so bot can expand beyond IRC. It's now possible to query TML over UDP socket directly, which allows us to run TML as a database or as a backend for execution of TML programs
- Continued work on serialization and persistence of Binary Decision Diagrams (BDDs) and TML runtime to allow restarting of a TML database backend without a need to recompress stored data through a textual form which is time and space consuming. Also usable but without memory mapping
- Created test runner script for regression testing so we can ensure old features are not broken by new changes.
Andrei
- Continued work on TML interaction playground relating to how discussions over Alpha client may look like. Showcases how discussions and code work around 2:00 in the video
- Discussion tree with mention counters implemented so we can track what's the most frequently mentioned entity in the discussion
- Will focus on TML next month
- Received friend request from Asen Lalov
Juan
- Continued on the arithmetic features. Wrapping up the features addressed over the last couple of months
- Completing two fundamental sets of operators: General operators (addition, mutiliplication, shift left and shift right) and Over BDDs operators which involved taking two sets of symbols and doing pairwise operations in only those two sets described last month. Successfully tested addition and multiplication as well as bitwise OR and bitwise XOR operators. We just need to complete bitwise NOT set of operators
- Also focusing on improving performance and implementation. Also with the help of Ohad, have been analyzing the theoretical combinations of arithmetic operators and its fundamental operations to reach the ideal implementation.
Fola
- Branding: we have a couple of designers working on a redesign of the logo and website. We're excited with the designs and can't wait to show them to the community once they're finalized
- Exchanges: two routes to get on reputable exchanges, either we have a huge community or by first getting listed on smaller exchanges and working up to bigger exchanges. Exchanges ask what other exchanges you're listed on, so for example, to get onto Gate you need to be listed on HitBTC, and to be listed on HitBTC you need to be on Whitebit etc. Which is why we've listed on Whitebit. We really want community feedback on whether you want us to be listed on HitBTC. We value your input here
- US exchanges are generally more valuable but also more difficult to get on and will not happen in the immediate future
- We also have people with good relationships with exchanges working with us (not sure audio unclear)
- Do check out Whitebit, we're had good communication with them so far. Check it out but of course as with any exchange, be cautious with your funds and don't leave them on there longer than necessary
- Hirings are taking a little longer than anticipated. Looking at an R+D position right now, and Umar will be joining in an official capacity soon, and looking for a researcher to work with Ohad
- Whitepaper is still being worked on
Ohad
- Dragan finished the difficult challenge of dynamic bit ordering of BDDs. Still need testing and enforcing strong types. Now you will need to associate a type with each variable in TML and thus we'll know how many bits to allocate for each type, potentially saving us exponential amount of time and space. This is a fundamentally big change.
- I spent a lot of time studying some rare literature that is very relevant to what we do, especially in the field if Boolean algebra. Basically they're the best methods of logical calculations derived by Bool himself 150 years ago.
- Also been trying to discover how to efficiently and soundly support compound terms. One way is to approach this from the world of monodic second order logic and view this as two relations; one of prefix-order equivalence relations and by manipulating these relations you can in turn manipulate compound terms. Another way is to use a specific data structure for terms (strings) which borrow tricks from BDDs. Another way is to enhance BDDs. We can state the problem like this: as you know BDDs represent a set of bit strings (0s and 1s), but we would like to support a set of bit strings that may be arbitrarily grouped with parenthesis in a way that has to align with unification. This also lead me to consider the problem of homomorphism between structures, trying to solve it inside our framework.
- As you know TML represents data in BDDs, well apparently there's another level below BDDs of data abstraction which is representing them as Boolean equations. BDDs can be seen as explicit solutions to Boolean equations but before solving them we can state the data in an even more succinct level than it is now. Achieving this in TML will not only lead to increase performance but also the ability to express uncertain data, among other benefits. Will also need to build a library for Boolean equations which will allow us access to more features including second order logic.
https://youtu.be/oYRmJqhLp5g
Includes QandA as usual (QandA text summary can be found on our subreddit)
Tomas
- Adding features which enables a rewrite of the Prototype bot
- Initial networking features added so bot can expand beyond IRC. It's now possible to query TML over UDP socket directly, which allows us to run TML as a database or as a backend for execution of TML programs
- Continued work on serialization and persistence of Binary Decision Diagrams (BDDs) and TML runtime to allow restarting of a TML database backend without a need to recompress stored data through a textual form which is time and space consuming. Also usable but without memory mapping
- Created test runner script for regression testing so we can ensure old features are not broken by new changes.
Andrei
- Continued work on TML interaction playground relating to how discussions over Alpha client may look like. Showcases how discussions and code work around 2:00 in the video
- Discussion tree with mention counters implemented so we can track what's the most frequently mentioned entity in the discussion
- Will focus on TML next month
- Received friend request from Asen Lalov
Juan
- Continued on the arithmetic features. Wrapping up the features addressed over the last couple of months
- Completing two fundamental sets of operators: General operators (addition, mutiliplication, shift left and shift right) and Over BDDs operators which involved taking two sets of symbols and doing pairwise operations in only those two sets described last month. Successfully tested addition and multiplication as well as bitwise OR and bitwise XOR operators. We just need to complete bitwise NOT set of operators
- Also focusing on improving performance and implementation. Also with the help of Ohad, have been analyzing the theoretical combinations of arithmetic operators and its fundamental operations to reach the ideal implementation.
Fola
- Branding: we have a couple of designers working on a redesign of the logo and website. We're excited with the designs and can't wait to show them to the community once they're finalized
- Exchanges: two routes to get on reputable exchanges, either we have a huge community or by first getting listed on smaller exchanges and working up to bigger exchanges. Exchanges ask what other exchanges you're listed on, so for example, to get onto Gate you need to be listed on HitBTC, and to be listed on HitBTC you need to be on Whitebit etc. Which is why we've listed on Whitebit. We really want community feedback on whether you want us to be listed on HitBTC. We value your input here
- US exchanges are generally more valuable but also more difficult to get on and will not happen in the immediate future
- We also have people with good relationships with exchanges working with us (not sure audio unclear)
- Do check out Whitebit, we're had good communication with them so far. Check it out but of course as with any exchange, be cautious with your funds and don't leave them on there longer than necessary
- Hirings are taking a little longer than anticipated. Looking at an R+D position right now, and Umar will be joining in an official capacity soon, and looking for a researcher to work with Ohad
- Whitepaper is still being worked on
Ohad
- Dragan finished the difficult challenge of dynamic bit ordering of BDDs. Still need testing and enforcing strong types. Now you will need to associate a type with each variable in TML and thus we'll know how many bits to allocate for each type, potentially saving us exponential amount of time and space. This is a fundamentally big change.
- I spent a lot of time studying some rare literature that is very relevant to what we do, especially in the field if Boolean algebra. Basically they're the best methods of logical calculations derived by Bool himself 150 years ago.
- Also been trying to discover how to efficiently and soundly support compound terms. One way is to approach this from the world of monodic second order logic and view this as two relations; one of prefix-order equivalence relations and by manipulating these relations you can in turn manipulate compound terms. Another way is to use a specific data structure for terms (strings) which borrow tricks from BDDs. Another way is to enhance BDDs. We can state the problem like this: as you know BDDs represent a set of bit strings (0s and 1s), but we would like to support a set of bit strings that may be arbitrarily grouped with parenthesis in a way that has to align with unification. This also lead me to consider the problem of homomorphism between structures, trying to solve it inside our framework.
- As you know TML represents data in BDDs, well apparently there's another level below BDDs of data abstraction which is representing them as Boolean equations. BDDs can be seen as explicit solutions to Boolean equations but before solving them we can state the data in an even more succinct level than it is now. Achieving this in TML will not only lead to increase performance but also the ability to express uncertain data, among other benefits. Will also need to build a library for Boolean equations which will allow us access to more features including second order logic.