/
2017-08-31 Meeting notes

2017-08-31 Meeting notes

Owned by Dennis Wisnosky
Last updated: Sep 07, 2017

Date

Attendees



Agenda


1) Where we are on our road map. 

2) Open Action Items

3) JIRA Issues Review - https://jira.edmcouncil.org/projects/RDFKIT/issues

4) Todays content discussion.



MB's About File work: 

https://wiki.edmcouncil.org/display/FIBORDFKIT/Product+Family)

https://wiki.edmcouncil.org/display/FPT/Ontology+Annotations+and+Metadata

5) For next week.

Proceedings: 

20170831 FIBO FPT RDF ToolKit

 

Agenda:  Begin with Action list

 

Glossary - headless thing.  DA stuck on a bug. Working with MD.   Sarawut Wariturnshit from Thailand is the support person. Or similar.  Progress is slow due to overnight responses.  JL: There is a support person in the US who may be able to help or liaise.   Maybe JG can help: issue is when the MD command line tries to contact the Teamwork Cloud. Is there something about our network set-up that might block connection attempts.  If so this might be why Sawarwut has been unable to help.   DA can run it when it does have a head.  Can also have the Head download things from the Teamwork Cloud, and can telnet to an address on that. Could also spoof a network client if that would help  JL: may be a problem with having a virtual X server.  Can look at the log and see if it's hung as a result of an aversion to that.  DA: Seeing a timeout. Invoke and wait is the thing at the top of the stack, which would be consistent with a timeout. This would also be consistent with the theory about the virtual X server (works OK with a real one). Might be waiting for that X server to answer.  JL: diagnostics - run the command line headfull.   That worked,  JL that means this is not a MD problem but would be an X server problem.   DA: something said that if your machine doesn't have specific X things working it might not work. Might give that a try.  JL: Try a different X client running it headless, and see if that hangs.  e.g. configure a virtual X server, try and connect an X Term to that and see if it hangs.  MD doesn't know if you are configuring a virtual X server. You just run the script. Just a black box, which doesn't try to open anything on the X server as far as DA knows.  JL: it will be opening such a thing. When you run headless the X client is running to localhost and may be trying to connect to such a client. DA would not know if it is doing this.   DA how it works headless: there is a script called Generate. Run this from the CL. This goes Starting Magic Draw, then does things opaquely, then tells you it's generated a report. Does not really look headless it just looks like a Command Line. DA has no idea what it is doing.  JL it is probably written in Bash.  DA it is in Bash and sets up things.  JL it probably also sets up a display thing. If you set it to talk to the display at a particular place, this would hang in the way described.   DA: In the CL you give it the connection to TWC, it goes off to there, it doesn't connect to your local MagicDraw.   Another way, connecting to MagicDraw that is local, (as suggested by Sarawut) works Ok.

 

DA recap: when I run the script to run MD headless, works on own Windows machine but when I run it on the AWS machine it gets a timeout problem when trying to reach the TWC.  So, is there anything in the network config of that machine that might stop it reaching out to the TCW server to do stuff? Telnet works though.   JG: Default firewall for all those boxes is inside out traffic always works but outside in traffic does not. So if it requires some messages in both directions it would fail.   What about loopback (local Host)? That will always work.  DA the script doesn't mention anything about X or about Display in general.   What connection?  Assume using a TWC specific protocol.

17  Is it http? ]

Dean Allemang

 

#!/bin/sh

# Add below code for fixing "Could not find or load main class com.nomagic.magicdraw.magicreport.CommandLine" of Mac OSX with Java 1.8

BASE=$(dirname "$0")

ARGS=""

 

if [ $# -gt 1 ]; then

                  while [ $# -ge 1 ]; do

                                    case "$1" in

                                                      -[a-z]*) ARGS="$ARGS $1" ;;

                                                      *) ARGS="$ARGS \"$1\"" ;;

                                    esac

                                    shift

                  done

fi

 

CLASSPATH=./reportwizard_api.jar:./reportwizard.jar:./lib/magicreport.jar:./requiredlibs/*

JAVA_COMMAND=java

EXEC_COMMAND="cd \"$BASE\" && $JAVA_COMMAND -cp $CLASSPATH com.nomagic.magicdraw.magicreport.CommandLineStartupChecker $ARGS"

eval $EXEC_COMMAND

 

if [ $? -eq 20 ]; then

                  # Two levels of indirection are here to preserve dollar sign

                  CLASSPATH=.:./requiredlibs/*

                  JAVA_COMMAND=java

                  MD_CLASS_PATH='$java.class.path'

                  MD_CLASS_PATH_I='$MD_CLASS_PATH'

                  VM_OPTIONS="-Xmx2000M -XX:PermSize=60M -XX:MaxPermSize=200M –

Dcom.nomagic.magicdraw.launcher=com.nomagic.magicdraw.commandline.CommandLineActionLauncher -Dcom.nomagic.magicdraw.commandline.action=com.nomagic.magicdraw.magicreport.CommandLineActi

 

DA puts script into skype window for JL

 

JG would like to also see the application.com file in case any port numbers are mentioned.  Uses a thing called AKKA that JG recognizes.

 

JG: try changing the debug level. Log in with SSH and run this on the CL.  Or run the original script?

 

ACTION:  JG and DA to meet 1:1 Sept 1.

 

Actions continued...

 

Next DA action = About to All files - Not done.  RDF Toolkit feature for catalog for developers.  Available in script form. Needed for RDFKit itself going forward but not immediately needed.  So close this action.

 

Artifact Ontology - Conversation is documented on the wiki (MB's notes about content of the actual ontologies).   EAL: Need to discuss both the Artifact Ontology and the Product Ontology.  EAL recommends we call the earlier thing the Product Ontology, as it contains a lists of the products we have built.  

 

What is next step - wiki, or discuss today?  JG at least discuss the generic requirements for the Artifacts ontology, on the RDFKit wiki site.  JG also rename this from RDFKit as it touches on a broader domain, on publishing of a giant ontology architecture and the governance around that.  JG do we all agree this is a separate product of the EDM Council,. that others can chip in publicly?  DW we decided this was open and public a long time ago.

 

Are there 2 ontologies (Product and Artifact) or one?  This is a packaging question. Let's have all the concepts defined on the wiki (Product, Family etc.) without having to commit to how these are packaged into ontology files.  For instance, might have a Module ontology, a Business Domain taxonomy, a business capability taxonomy and so on.  BIAN has already created a banking business capability taxonomy - we can OWLify that.

 

What do we do now?  Create a working Artifact ontology, and start using that.   DA: we are already doing that to some extent.  DA one issue (PR can work on) is that what we are calling Artifact Ontology or Product Ontology, the Specification Metadata (SM from OMG) has a lot of similar goals. A lot of the things are already in there. So there is a need for co-ordination. Was released through OMG but how do we do governance of the things we want to use now?

 

DA would prefer this is something we build, whether or not the OMG uses.  EAL: We are packaging an artifact ontology and spec metadata. The part where we describe the metadata we want to attach at a concept level, ontology level and release level, which joins work done by OMG. Then there is where we describe the actual artifacts (products) that the Council is producing from a technical perspective, from which we build the RDF Toolkit.  Then there is a part where we describe the EDM Council landscape and products, with focus on FIBO Products.  EAL has worked on the latter of the above.

 

EAL suggest that before PR and he go further, need to articulate these into detailed requirements. As per above JG suggestion (Personas, user stories etc.)  JG: Let's start writing this down.  https://wiki.edmcouncil.org/display/FIBORDFKIT/Concepts

 

User stories can fall out of the process definitions MB has started to work on in the Process page, which is FIBO specific and not public. These become one source of user stories.  DA at each step, where is the script doing something bespoke that can be model driven.

 

ACTION:  EAL download the latest FIBO Infra repo from GitHub, find fibopublish.sh and read the comments. Identify what can be modeled.  Bring questions from that to the next meeting.  https://github.com/edmcouncil/fibo-infra/blob/master/jenkins/bin/publish-fibo.sh

 

JG demonstrates a user story. Has Personas, Features and Outcomes. Also flags. Once a US is well defined (Ready) then link to a JIRA issue, give it an owner, then do things like translate to SPARQL and so on, to demonstrate how it worked.   MB one such business user is whoever would make the decisions that go into published metadata.   EAL is not clear how this relates to the work he is doing. Suspect this is a whole separate thing.  PR: this whole thing around Modules, Specifications is certainly something that needs thought.  PR - can help co-ordinate. Also the earlier MB proposal for what the actual Business Domains is part of this conversion.

 

PR to EAL: Product v Artifact ontologies. One would be the things we develop (Modules, domains etc.) and the other is the Products (OWL, SKOS etc.)  PR: Are we trying to develop both aspects, the development aspects (how we structure the ontologies (review our modularization)) and the other aspect?

 

DW focus on the short term for now. Put the Development (modularization) aspects on next FLT?   JG: Modular structure versus business domains view.   EAL: File structure is also relevant to this. That file structure and whether it corresponds to Module.

 

MB These questions are separate from and orthogonal to the Products (deliverables, OWL, SKOS etc.) matters.  MB, OPR, EAL and JG all agree that these are orthogonal matters.

 

PR this would also involve creating routes through the content.   One ontology can only be one module but may be related to multiple business domains.   All agree.

 

DA: did the Headless thing work?  DA wants to visit additional debug info that he has send something about to JG.

 

Next Action: OK -  Fragment Server. Omar is going to pick that back up, grab whatever is published, pick up that task for the Sept 30 deliverable.  Action description is fine as written.

 

Action: JG and TC to get together on making TBS work. That happened. And JG has fixed the thing in question. TC is happy.  TC: Now there is a problem on my side with an error in the build, Can resolve that. Now will be clear to do builds.

 

Landing page for the OWL (DA, JG).  Not done yet. DA and JG both busy on other things

 

JG fixed an 403 error on nGnix last week. DA didn't know.

 

INFRA 191 -  for VOWL.   Assume we will provide VOWL as a deliverable in our Sept 30 Release.   JG has a colleague in India who did this for BNYM and would like to do this for EDM Council.   JG to proceed with giving him the rights to do this?

 

Next action DW has seen the generic picture of what is running what from JG.  JG has action to customize it for FIBO.

 

DW:  Concept is a word for which more than one concept seems to apply. Need to make the language in the FIBO Primer sharper to accommodate the range of usages of this word.

 

DW: Next week we start to meet 2x a week again, Tuesday 10am and Thursday 11am EDT

Decisions:

Action items