Okay so make Mind-Agents managed their own resources. Yea that makes sense. You'd still need some simple Agent-Manger but yea I can see how that would clean up the overall design.
Thanks for clearing my Mind. It got a bit stuck after redesigning ECAN and depending on the CogServer so much :) On Friday, June 9, 2017 at 12:08:23 AM UTC+2, linas wrote: > > > > On Thu, Jun 8, 2017 at 2:22 AM, Roman Treutlein <[email protected] > <javascript:>> wrote: > >> Hey, >> >> so this is mainly a question to linas. You seem to want to get rid of the >> CogServer and I was wondering what exactly you would prefer in its place. >> The way I see it the CogServer currently serves a vital role in the whole >> Architecture by providing a way for all the different parts of OpenCog to >> come together in one place. >> And by managing a bunch of threads used by the various modules. >> > > Except it does none of these things. You can use opencog, via both python > and scheme, without ever starting the cogserver. You can use both python > and guile together, even, without starting the cogserver. > > Now, the cogserver provides a network interface, but so does guile, and I > presume that there's some generic python module out there that would do > this too. So you can start the a scheme or a python network server, and > remotely log into opencog, without ever actually starting the cogserver. > So basically, the cogserver isn't needed for neworking. > > I'll bet Haskel also has some network server module, too. I'll bet that at > least some of these network modules might be pretty fancy, providing > encryption, login credentials, other security bells and whistles that we > sure-as-shootin are never going to add to the cogserver. And we shouldn't > -- it breaks the concept of modular design. If some module out there > provides secure, safe, audited, encrypted network access, we should use > that. We should not reinvent the wheel, badly, (which is what we've done) > by putting this function in the cogserver. We should remove this > mis-feature, not enhance it. > > So what else does the cogserver do? Well, it manages "mind-agents". Except > (1) the code for managing mind-agents is badly implemented and completely > buggy, and (2) you do NOT need a cogserver to manage a pool of mindagents! > These are completely distinct, unrelated concepts that, due to historical > accident, got mashed together into one. > > If the mind-agent code was redesigned so that it worked correctly, and so > that it was a stand-alone module, instead of a lean-to, then the cogserver > would become just a network server, and nothing more, and, as I say, there > are surely better network servers out there. The cogserver is currently a > crutch, a hack job. > > > >> >> And those seem to be functions that we will always need. So I don't >> really see how the CogServer could be replaced. Improved and Changed sure >> but not replaced. >> >> So how do you see OpenCog working without a CogServer? >> > > Well, easy. I do this every day, all the time. I only rarely start the > cogserver. So, for example, in guile > > $ guile > guile> (use-modules (opencog)) > guile> (use-modules (system repl server)) > guile> (call-with-new-thread run-server) > > Then in another shell: > telnet localhost 37146 > that's the default port nummber. To change it, see > https://www.gnu.org/software/guile/manual/html_node/REPL-Servers.html > > then guile> (Concept "foo") > > Bingo! -- you've just used the atomspace -- logging in remotely, using a > two-line replacement for the cogserver. One line to load the network > module, one line to start the network module. What do we need a cogserver > for, again? > > I'm sure you can get something similar in python and heskel. > > --linas > > >> >> best regards >> >> Roman >> >> -- >> 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] <javascript:>. >> To post to this group, send email to [email protected] >> <javascript:>. >> 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/fb65213f-7bb7-4746-9893-1d4190f2474c%40googlegroups.com >> >> <https://groups.google.com/d/msgid/opencog/fb65213f-7bb7-4746-9893-1d4190f2474c%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/0de52479-66ee-4ece-928d-90898f1a3abd%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
