Bitcoin Forum

Whilst there is a lot of excitement about the possibilites of "Turing complete" blockchain transactions that are supported by Ethereum and AT (the latter being my own invention) after studying how these work I have come to the conclusion that they are just not going to really scale up to provide us with a "new internet" (which is what I have now come to think of as being the main goal that blockchains should be used for).

If you think of how things like this forum work today (via HTTP/HTTPS) you create a forum post by effectively populating a form with values that gets turned into a HTTP/HTTPS POST request for the server to then turn into a DB operation (after performing validation).

But both Ethereum and AT instead are actually working more like low-level machines (either a Java VM in the case of Ethereum or a virtual CPU in the case of AT) which are very inefficient in comparison to something like HTTP with a web server backend that uses say FCGI.

They have to work this way as otherwise you could create a very simple program like this following:


that will run your computer out of memory (causing it to either crash or become completely unresponsive) unless the fees to execute more than X amount of steps are high enough.

But is there another way?

Yes - I designed a concept called Software Manufacturing (http://ciyam.org/open/?cmd=view&data=20121221010507352000_P&ident=M100V112&chksum=b3d538bf) many years ago (which works in a somewhat similar manner to Charles Simonyi's project called "Intentional Software").

In Software Manufacturing the above pseudo code would have to be divided into two components - a looping specification and a separate specification that covers what to do within the loop. As all source code is "generated" you can't stop the resulting code from effectively being something like this:


This is not something you can do using either Ethereum or AT as they are simply not high level enough but this is something that can be easily be done using Software Manufacturing (as you have no choice).

Using such a high level approach means that rather than bothering each node with putting a VM or virtual CPU between the network and the DB you can just put in the functional operations themselves (which would look more like SQL statements than machine code).

This will scale to at least the level we have today in regards to internet applications and this is what CIYAM is currently developing.

If you'd like to learn a bit more about what exactly Software Manufacting is then a quick introduction can be found here: http://ciyam.org/docs/methodology.html.

While I appreciate the project and love Vitalik's research, Ethereum appears to be an act of supererogation with the burden of added inefficiencies which could be carried out by one or multiple oracles.

http://gavintech.blogspot.com/2014/06/bit-thereum.html

and even if Ethereum would be a better system, they dont get traction.

it would be much better if they would work with BTC instead of this project  :-\

Yes - that I think is an important point - CIYAM is not about trying to create a new "currency" so in fact it will be able to work with Bitcoin (there is already a Wallet package).

The vision that CIYAM has for blockchains is for a new internet but there is no reason that Bitcoin shouldn't be the main "store of value" (or general financial transaction mechanism) in the new internet.

Shameless plug / Product placement

AT is an open source and free product - so okay call it a plug but I don't make a cent from anyone using it (and it is in fact the first Turing complete system to be running on a mainnet).

And why should I be ashamed of it?

MaidSafe will lead the way into true decentralization.

I think it is another great idea - it is probably just going to be a question of which works best.

Whether you make a profit is irrelevant. General Discussion is a place to discuss about Bitcoin, not to promote other projects. Please post in alt-currencies.

CIYAM is not an alt currency (it isn't even a currency) and it has a Bitcoin Wallet so please get your facts straight.

Why do you think it is a product placement from the moment the thread name already contains that word and on top of that he is saying that is not the future. He is not advertising it this way, just discussing his thoughts and concerns. And even so -  as CIYAM mention it is the first Turing complete that is live on mainnet, so it is worth mentioning imo.

Lumping together a random project with a blockbuster such a Ethereum and taking a critical tone to make the post sound legit may be smart marketing, but it's still completely off-topic in this sub-forum.

That being said, Satoshi and I could be wrong sometime, as being pure rockstar geniuses doesn't necessarily make us fit to debate of political matters; now Satoshi has stepped down, so that let's only me to rule that forum. Right ;)?

That is rather odd to say - as in fact the CIYAM project was started in 2001 (yes 14 years ago) but okay seemingly you are in charge of what is "on topic" in the whole forum now.

If I was I would have moved your thread to alt-currencies. Now I'm not saying that your project isn't interesting. I'm just stating that you are off-topic here, in a subtle way maybe, but still off-topic. I let it up to the mods to make the final call on that matter.

So again it seems that you think you should be in charge of what posts go where.

Like you say - let's leave it up to the mods. :)

(as you think that a product that has a BTC Wallet and isn't an alt coin should be moved to alt coins)

Have you read the documentation of AT or studied its potential to say it is a random project and/or cannot be compared to Ethereum? We are not care about marketing or whatsoever, just the tech and this is how that random project was developed. It is not about being wrong or genius or whatsoever, anyone is free to express his opinion and is also free to have a huge ego. But if you support the evolution of the technology and moving forward, at least you should be able to have a serious discussion instead of "trolling" and accusing. If the thread is or not in the right sub-forum, then are mods out there to decide that.

But lets not speak more about AT. The thread is not about AT or Ethereum. The concerns raised from CIYAM are beyond these projects and are more general.

I think he is just making an emotional knee jerk reaction because you had the nerve to attack his investment in ethereum. Of course this topic is relevant to Bitcoin as it makes some salient points that more efficient processes are best handled off the blockchain.

+1

*sigh*

Expected a nice intelligent discussion starting with 18 posts already.....instead found trolls and whiners.....that is enough BTT for one day (all 5 mins of it).

Please elaborate your post and show how you propose to linearize your "software manufacturing" representation of this sample program.  It has to have a one-dimensional representation to be really useful.

All I saw in your links is this picture which (I cannot even post the link, because your Javascript & DOM are so complex that they completely don't render in Safari and in Opera I can't "Copy Image Address"). I have to transcribe the portion of the image:


If this is another attempt at "visual programming" then it will probably fail to provide progress in the same way as all previous attempts at two-dimensional programming degenerated into "freehand paint programs".

Okay - it is not an attempt at "visual programming" at all but works more using "aspect oriented programming" methods.

I can explain the details to you if you want (it isn't that hard to follow).

Yes - but you are welcome to help further the topic with something more interesting (am sorry that I seem to attract a lot of trolls these days).

Sounds very much like functional programming with a recursive combinator to handle the looping using tail recursion and payload passed a lambda. Are you basing your work on some existing funtional language or starting from scratch?

JIT compilation? Otherwise how do you deal with differences of architecture beween hosts?

I'm a bit unclear about what you are proposing. My understanding so far is that you are working on a contract system using pure functional programming and just in time compilation to optimize the code so that it runs faster than a VM or interpreted stack based language and you count leveraging the nature of functional programming as side-effect free modular function objects to allow for something completely modular, and are planning on pushing that to the extreme where one would just need to connect functional boxes to do any type of program without having to actual write a program. Is that correct?

PS: ok, that doesn't belong to alt-currencies, but I still think you should have this thread moved to projects. You'll get more attention from developers, and a better signal / noise ratio.

If the mods see fit to move it to Projects then that is fine with me (it is often hard to know where best to start such a topic so I dropped in the Bitcoin Discussion forum to see where it would go).

It is not functional programming although it does work close to that (Aspect Oriented Programming is probably the best analogy).

It doesn't use anything like JIT as it actually generates C++ source code that is compiled.

The efficiency is that the information passed is very minimal (think like SQL) so it can do a lot more processing with a lot less input.

Do you have a spec of your language?

LLVM? YACC parser generating C++ at run-time and calling the compiler?

You mentionned earlier that your project is open-source. Did you already commit the code somewhere?

It is not a language (in any traditional sense) so that might take some time to explain (but it generates standard C++).

Creating the documentation is a going to be my main task in the future - so please be patient with that (if you want to contact me more directly about the project then feel free to send me a PM).


The project is here: https://github.com/ciyam/ciyam

Hey don't worry, I attract them just the same :)

We're doing similar Turing complete development here with eMunie and have had many a discussion of how to solve problems just as

while (1==1)
{ str += str; }

Our interpreters run a Java/JavaScript hybrid, the bulk of the script code being JavaScript, with calls possible into native Java functions depending on the permissions granted to that script by the executor.

Side stepping a lot of the technicalities, we investigated a number of ways to detect this behavior but couldn't come to a accepted method which allowed legitimate scripts to operate as intended, but malicious scripts to be detected (which was as we expected with a Turing complete implementation, but wanted to check it out anyway).  

Ultimately we have decided on an implementation which doesn't attempt to combat it directly via outright detection, but allows the script itself to provide meta-data which indicates how much memory it is likely to use, average time between heartbeat signals the interpreter can expect and various other metrics.  The creator of the script can execute a benchmarker which generates this meta-data, before deployment, and packages it into the script header.

With the parameters set, the executing nodes then have some information on expected behavior and requirements to execute.  If the requirements are too great for that node, it can choose to not execute that script (Raspberry's for example), or pass it to the network for execution and receive a result, as similar to Hadoop Map-Reduce architecture.  Additionally if the execution of that script exceeds the expectations, perhaps the script management module doesnt receive a heartbeat from that interpreter instance for a period of time beyond the expected average as per the meta data, that script can be terminated as each interpreter instance runs in its own sandboxed VM instance.

This kind of implementation should suffice, as the issuer of the script has to provide some information on what the script requires, if it exceeds those expectations by being malicious, then its very easy to detect and act accordingly.

Parser is handwritten!? Do you have the grammar specification of the input and/or example programs?
Is the code on Github functional already?

Your site mentions that CIYAM is a linux distro but what you descibe here seems more like some sort of runtime environment.
Are you booting directly into the runtime env or are these two separate projects with the same name?

That is an interesting approach - the way that Software Manufacturing can solve the same problem is actually must easier (because the source code is *controlled* by the generation).

So in a CIYAM system we don't need to do these checks (although of course you do need to check the app itself before you decide to run it).

All the code is there - the main "parser" you might be interested in is called "xrep" (there are actually a few parsers).


The CIYAM Server OS is going to be the way to distribute CIYAM (my view is that VMs are the future).

Well, the spec of requirements for our scripting solution, and architecture, is quite different to what is currently being developed such as Ethereum and others (AT too no doubt).  We have multiple execution domains/scopes, where the requirements for each domain may differ between them for the same script/s...so the solution I present is very tuned to our needs.

I'm sure that with a lesser/looser set of solution requirements it could be implemented many ways, perhaps simpler.

We are keen to work with any project that is furthering the concept of blockchains so if you think CIYAM can help then feel free to contact us.

Unfortunately we left blockchains behind around 2 years ago.  What about channeled ledgers which are essentially many many distributed transaction trees? ;)

CIYAM will let people create their own blockchains and their own apps on top of it. So perhaps we are not so different.

Market doesn't think so - http://coinmarketcap.com/assets/maidsafecoin/#charts (switch into BTC valuation to get rid of Bitcoin price influence).

Thanks. I'm not clear looking at the grammar how you specify loops. You have stuff to handle sets and running functions on sets, but what of the cases where you don't have a set? You create a dummy set just so that you can iterate on it? Maybe if you can point me to a good example of some basic loop which test something and quit when the test fails (like you gave in the first post), it will help me figure what in the language is to be interpreted as a loop.

You answered to 2112 that you aren't using visual programming to create the programs but you also said that the paradigm doesn't require any actual programming. But looking at your example files, xrep still appears to be a form of script. Do you mean to say that the "developer" won't have to write anything in xrep to make applications? If so, in what form will the developer "develop" his application?

That's an interesting approach. If you ship the solution as a bootable iso, you have solved at the same time the problem of maintaining compatibility accross hosts since the host is standard so less work to make your make / autoconf and making sure the thing stays functional accross distribs and versions of linux, and also solved the problem of having to trust that the OS itself is secure (which it isn't in the overwhelming majority of cases). It also easier to run the thing as a VM.

Other than price / market feedback (markets aren't efficient, specially on the cryptoscene), do you have any other reason to believe that MaidSafe isn't going to meet their objectives?

Meet objectives != Lead the way. I think their chance to meet objectives is quite high.

Perhaps a good start would be to reimplement Bitcoin over your plateform.
Easier said than done, but without that as an exampe, I doubt anyone less knowledgeable of your system would manage to build a whole blockchain based application.

When you say that people can develop their own blockchain, do you mean it in the Bitcoin sense of an actual chained list of blocks, or in the more general sense of a general datastructure  / container with cryptographic operations to enforce determinically precedence of concurrent modifications?

Another question regarding extensibility: do you plan to allow interfacing the high level code with custom modules written in a lower level language (C/C++ for instance)? do you plan to provide some GUI building toolkit usable directly from within your high level language?

Many questions. The best would be to have some presentation page and tutorial pages that cover the concept, features and distinctive advantages of CIYAM so that people can make their own idea of whether they would have a use for that. Actually, for a 14-year project, I'm a bit surprised that there isn't much more documentation.

If that's not asking too much, in what context did you develop this framework? Do you have a particular business model / pricing / licensing for the software or are you providing that totally free without any strings attached?

And who leads the way? NXT ;p?

Hard to say, there are so many decentralized technologies. A new pops up every day.

Thanks, but I don't want to read prose explaining things. I wanted to see the textual representation of your "software manufacturing" rewrite of the short buggy C++ snippet and the safety-improved C++ snippet. AOP is a know paradigm and I don't see a point of re-explaining it.

I just wanted to be able to se qualify how much technology is in what you propose and how much is pure marketing bullshit. Your "methodology" link is pretty much own-words rewrite of the "Microsoft Site Server 3.0 Commerce Edition" marketing blurb. Did you intentionally plagiarize it (counting on the readership not remembering Microsoft Windows NT 4) or was it because your life experience was more or less equivalent to Microsoft state-of-mind circa 1999 (with C++ substituted for Microsoft Visual Basic)?

Point of though for you: Chess is essentially a two-dimensional game yet chess players developed linear one-dimensional  representation of chess games to be able to talk and write about it.

http://en.wikipedia.org/wiki/Holographic_principle

http://en.wikipedia.org/wiki/Not_even_wrong

In regards to the "marketing blurb" (and I'll use that term lightly as the one thing I am hopeless at is marketing) it was not copied from anything else (and I am not even familiar with the Microsoft product mentioned).

In regards to loops there are two specifications: for_each and for_each_record with the second being based upon DB records.

To see how CIYAM's AOP implementation works you can just take a look at some of the Meta*.cpp files (you might not want to look at Meta_Specification.cpp though as it is an extremely large file).

To be clear CIYAM's generated software *is* C++ but you don't use C++ to write it - you use an application (that looks a bit like http://ciyam.org/open but called Meta). As you are unable to "write actual source code" you are unable to prevent it from writing "code that includes whatever checks are deemed necessary" (and in fact the checks could occur at other key parts of the generated code so may not have to appear exactly like the string appending loop example).

The AOP approach uses the *.spec.xrep "snippet" templates which are fed into ciyam_class.cpp.xrep (which itself is created from ciyam_class.cpp.xrep.outline so you'll only find the latter in the repo). The makefile implementation is also constructed using xrep and another tool (so there is a specific makefile.tools to act as a bootstrap and as Meta is also written in CIYAM there is a bootstrap console app version to build it).

The technology platform has been used commercially since 2007 (I created the open source project in late 2012) and the CIYAM Open website was created using this platform (on a Galaxy S3 in roughly 5 minutes).

What hasn't been finished yet is the blockchain implementation (which is why you won't find any docs on that yet) but that is mostly done (perhaps a month or two away from being ready to launch).

The CIYAM Server (which should be installed as a VM) can be downloaded from here: https://susestudio.com/a/kp8B3G/ciyam-server

Expect quite a few updates in the following weeks and months.

Anyone interested in the project can follow the progress from here: https://bitcointalk.org/index.php?topic=598860.0

Why is this a problem? Since the script is running in a VM, it does not matter if the VM runs out of memory, as the entity running the script is paying for the resources.

It isn't a problem where those resources have been very tightly controlled which is how Ethereum (and AT) work.

But such systems are very inefficient - so my point is that we are more likely to see much higher level approaches in the future (which is what the CIYAM project is about).

BTSX is interested except they diluted shares.

When is the project likely to be completed?

A project such as this is *never completed* but what I can say is that a working system can be expected in a few months.