sorry, just unearthed this email.

The diagram now looks correct, but I am concerned about emphasis:
everything interesting in opencog happens in the main two blocks -- the
fact that there just happens to be a compile-time dependency on python,
scheme, postgres, cogutils doesn't seem very important. I guess it should
be noted, but its not worth a lot of graphical real-estste.

Besides python and scheme, there are also c++ and haskell bindings, and
also a REST webapi, and also, a ZMQ binding.  The C++ and the REST bindings
are in active use. the haskell bindings are exprimental.  The ZMQ bindings
might have bitrotted.

Besides postgres, there is also gearman for distributed processing, and
also a binding to the java neo4j database via the zmq bindings.

Anyway, none of these things are part of the "architecture", per se; they
are all outside of the architecture, they are on the periphery, a way of
attaching to the achitecture.  The actual architecture is that of the
atomspace, and the 100 different atom types, the rule engine, PLN -- that's
the actual architecture.   These are really what make opencog work. The
fact that there's python of scheme of C++ mostly kind-of doesn't matter at
all.

--linas




On Sat, Oct 22, 2016 at 3:05 PM, Apil Tamang <[email protected]>
wrote:

> Hi Linas,
>
> Did a little bit of reading around along-side your points. Here's a
> diagram I kinda ended up with. It's definitely not exhaustive, but let me
> know if is fairly more accurate. As I understand more of opencog, I'll try
> to keep updating it.
>
>
> On Sunday, October 9, 2016 at 7:02:25 PM UTC-4, linas wrote:
>>
>> Hi Apil,
>>
>> Interesting start.  If these were fixed up a bt, they could be published
>> on the website.  As it is they are not quite right/misleading.
>>
>> So -- the atomspace depends on guile and python, so those arrows should
>> point at the atomspace.
>>
>> -- The reasoning agents depend on the atomspace (they are actually a
>> built-in part of the atomspace --  a subcomponent), they do not depend on
>> opencog, nor the cogserver.
>>
>> -- the  cogserver provides some low-level network access, it should not
>> be elevated to any special status.  Programmers are quasi-welcome to maybe
>> propose and use other network daemons/protocols, instead of the cogserver.
>>
>> -- the cogserver provides the opencog shell. That's the network interface
>> that it provides -- the shell is a subcomponent of the cogserver.  You do
>> not have to use the cogserver to get work done.
>>
>> -- the atomspace depends on zeromq, but only if you turn it on. It is one
>> of the alternative network stacks to the cogserver.
>>
>> -- Nothing that I know of depends on opengl, sdl or thread building
>> blocks  These are old hold-overs from some old visualization tools that
>> never worked well.  These should not appear in the diagram.
>>
>> -- nothing in opencog depends on moses.  It should be in the diagram, but
>> currently, its not attached to anything (except that it depends on cogutils)
>>
>> -- only the NLP subsystem depends on link-grammar.
>>
>> -- the atomspace depends on postgres, for save/restore from disk.  This
>> is important, and should appear in the diagram.
>>
>> Linas
>>
>>
>>
>>
>>
>> On Thu, Oct 6, 2016 at 8:58 PM, Apil Tamang <[email protected]>
>> wrote:
>>
>>> Hey All,
>>>
>>> So, spent some time researching on the OpenCog architecture. From the
>>> little given in the github/opencog <https://github.com/opencog/opencog>
>>> and github/opencog-docker <https://github.com/opencog/docker> links, I
>>> spawned some high level dependency diagrams of the various subsystems. I am
>>> having a bit of difficulty relating what goes here, but I hope the diagrams
>>> are relatively okay. Also, I'm having a bit of difficulty relating the
>>> docker files to the opencog subsystems.
>>>
>>> I've put some of my questions on the diagrams themselves. Could anyone
>>> take some time to look at it and respond?
>>>
>>> I think it would be a good idea to grow these diagrams in more detail
>>> and clarity. For a newcomer, it is otherwise quite hard to understand how
>>> everything fits into the place. I can work on it while I'm ramping up on
>>> the project.
>>>
>>> Is anyone willing to volunteer like an hour or something over
>>> skype/google-hangout to help me get some traction? There're just a ton of
>>> questions I have about getting setup.
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "opencog" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> To post to this group, send email to [email protected].
>>> Visit this group at https://groups.google.com/group/opencog.
>>> To view this discussion on the web visit https://groups.google.com/d/ms
>>> gid/opencog/a5e60576-b68c-48d2-bab3-7431f8eb2278%40googlegroups.com
>>> <https://groups.google.com/d/msgid/opencog/a5e60576-b68c-48d2-bab3-7431f8eb2278%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
>>

-- 
You received this message because you are subscribed to the Google Groups 
"opencog" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/opencog.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/opencog/CAHrUA352f5vaAjiCoKhbYggEVY%2B%2Bjv4NBXDma%3Di7SaDdwVmiWA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to