nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

FTA diagram: Javascript based probability

medini_1036016653

Dear Jan,

new week, new script... I'm currently evaluating the possibilities the JavaScript feature gives me to set an event's probability. I'm able to set the probability for example based on the event's name "E194", but I fail to traverse the model:

How can I access the parent?

Actually I want to access the other event of the AND gate and then use mediniGetOpposites() to first "navigate" to the system model and then continue to access the according DC worksheet (FMEDA) failure mode entry and extract the applied diagnostic coverage "spfCoverage".

While experimenting I noticed that the expression seems to be evaluated twice. I renamed the description from "1234" to "bla" in the diagram and then the output "Events name: bla" got displayed twice. I'm on 2019 R1.1.

Best regards

Peter

Find more posts tagged with

medini

medini legacy

Accepted answers

medini_1024169437

Hi Peter,

no problem, that's what the forum is good for :-)

For FTA scripting its important to know the difference between Event and EventNode, see the threat that i had with Ralph recently: https://medini.freshdesk.com/support/discussions/topics/1000093877

The expression is stored at the Event but the hierarchy (parents, children etc.) are at the EventNode.

I attached a script that converts MOEs to Transfer Gates so you get an impression.

I'll double check why the expression is evaluated multipel times, it might be not cached at all times.

Jan

All comments

medini_1024169437

Hi Peter,

no problem, that's what the forum is good for :-)

For FTA scripting its important to know the difference between Event and EventNode, see the threat that i had with Ralph recently: https://medini.freshdesk.com/support/discussions/topics/1000093877

The expression is stored at the Event but the hierarchy (parents, children etc.) are at the EventNode.

I attached a script that converts MOEs to Transfer Gates so you get an impression.

I'll double check why the expression is evaluated multipel times, it might be not cached at all times.

Jan

medini_1036016653

Dear Jan,

perfect! I missed that detail. Let me explore that path. It looks promising, in any case :-)

Peter

medini_1036016653

console.log("Event's parent name: " + event.nodes[0].parent.name);

yields the AND gate's name. Nice!

medini_1036016653

Jan, this works beautifully!

Unfortunately, calculating this minimal tree makes the JavaScript expression being executed five times. (I cleared the log prior to execution, doing a right click on E199 -> calculate probability):

Moving any single (unrelated! *) event on the diagram makes the expression being executed twice. I will bring this kind of script soon to several thousand nodes, so expect to have this *some* performance influence :-)

Best regards

Peter

(*) the unrelated element on the diagram is simply a newly created event with default settings, so does not contain any JavaScript expression itself.

medini_1036016653

Dear Jan,

so I intend to improve traceability and create a trace. It seems that it is not possible to load a helper script (factory.js) or at least I haven't guessed the correct "cwd" (current working directory) from those script's execution:

// create trace between FTA event and failure mode in FMEDA
// load(".lib/factory.js"); // FIXME: does not work

console.log(__filedir);

makes the script execution fail as well, I guess this might go beyond (initially) intendended functionality?

Best regards

Peter

medini_1024169437

Hi Peter,

thanks for the feedback, I guess the probability might be indeed called multiple times as it seems.

Lets see whether you face any "p-issue" (so performance problem).

You understanding the load() inside a probability script goes beyond original intention is true.

In fact that even would slow the script further down since file operation is required and load() does not cache the file content. Anyway, I can check whether it is supported or not.

Happy Scripting

Jan

medini_1036016653

Dear Jan,

so I went down this rabbit hole a little further and "reimplemented" (copied) key aspects of factory.js to that node's JS expression, in order to be able to create traces:

// $EXPERIMENTAL$ $ENHANCED_CONTAINMENT_ACCESS$
// $DEBUG$ <- append to first line to enable debugging
//load("scripts/.lib/factory.js"); // FIXME: does not work
// .. so do it the tough way and reimplement key aspects of "factory.js"
var TransactionUtil = bind("org.eclipse.emf.transaction", "org.eclipse.emf.transaction.util.TransactionUtil", false);
var CreateTraceOperation = bind("de.ikv.medini.kernel.traceability", "de.ikv.medini.kernel.traceability.operations.CreateTraceOperation", false);

 // helper
function __createTrace(from, to) {
 // create one
 var domain = TransactionUtil.getEditingDomain(from);
 var op = new CreateTraceOperation(domain, "Link", from, to, "");
 op.execute(null, null);
 return op.getTrace();
}

Unfortunately directly calling "__createTrace(from, to)" with two objects fails and leads to this debug console output (I start medini with the flag "-debug"):

Status ERROR: de.ikv.medini.scripting.js code=4 Failed to execute Java Script: Java constructor for "de.ikv.medini.kernel.traceability.operations.CreateTraceOperation" with arguments "de.ikv.medini.cockpit.gmf.edit.MediniDiagramEditingDomainFactory$MediniDiagramEditingDomain,string,de.ikv.medini.metamodel.FTA.impl.EventNodeImpl,org.eclipse.emf.common.util.BasicEList,string" not found. (unknown#17) org.mozilla.javascript.EvaluatorException: Java constructor for "de.ikv.medini.kernel.traceability.operations.CreateTraceOperation" with arguments "de.ikv.medini.cockpit.gmf.edit.MediniDiagramEditingDomainFactory$MediniDiagramEditingDomain,string,de.ikv.medini.metamodel.FTA.impl.EventNodeImpl,org.eclipse.emf.common.util.BasicEList,string" not found. (unknown#17)

I tried to debug this with the Rhino Debugger (yes, this actually works somewhat :-D, when instructing the pragma "$DEBUG$"), but couldn't see where this goes wrong, as my debug session get's interrupted by a new one after some seconds (is there some background task triggering a refresh of calculations?).

Do you have a clue?

My backup plan for traceability was to set the "represents" attribute, but unfortunately I'm only allowed to select a single object. In my case I want to map the event to 0..n other objects. hmmm...

Best regards

Peter

medini_1024169437

Peter,

I think "just" the second argument is wrong. The error states that you pass a list, but it actually should be a single object

so I assume your call is somewhat wrong.

Note: Apart from that I see some trouble in this approach, even if it may work technically.

You create a trace inside a scripted probability method which may called several times so you create traces while

the probability is computed? Sounds a bit strange and risky to be if you ask me :-)

Do you at last check whether that trace exists already?

Happy Scripting

Jan

medini_1036016653

Hey Jan,

well observed. Let me check, but I guess you are right!

And of course you are right with your cautious inquiry. It will not be wise to actually do that trace creation in this script. Furthermore, I actually haven't implemented a check for existance of that same trace yet either.

So wrap up: I reached my goal of programatically setting the probability to the diagnostic gap (one minus the diagnostic coverage (1-DC)), whereas the DC is set in a FMEDA - based on the actual (expert) judgement of a failure mode and NOT based on a SM's expected DC (as listed in a Safety Mechanism collection).

Thank you for your extrremely timely replies. The helped a lot!

medini_1024169437

You are welcome Peter.

Please mark an answer later if you can.

Happy Scripting

Jan

medini_1036016653

Done, although I can't mark any of my replies as "the answer". This is my solution for accessing the sibling event in my originally shown tree:

var FTA_FM = event.nodes[0].parent.children[0]; // sibling event, representing failure mode: nodeEvent's first node's parent; and parent's first child

Be aware that usage of "[0]" is fragile, but for now this works for me.

Happy scripting!

Peter

medini_1024169437

Thanks Peter for the summary, highly appreciated. True that accessing any node or sibling is fragile

but general approach should be clear.

Happy Scripting

Jan