Describe the bug Hey guys, I'm working with the grule engine and I came across the following situation.

I created a microservice that uses grule to analyze data over rules. In this microservice I have an API that allows adding and deleting a rule. But the service is crashing when I remove an existing rule and add a new rule. So, I created a test application only using grule to analyze the problem and I came across a panic: DEBU[0000] Cloning working memory POCRules:0.0.1 lib=grule-rule-engine package=ast DEBU[0000] Cloning 7 expressionSnapshotMap entries lib=grule-rule-engine package=ast panic: expression BMF.Value1 is not on the clone table

Apparently the problem is when removing the rule just on the ruleEntries but the workingMemory remains the same.

I will index the sample code that reproduce this error.

So my question is, Am I using rule removal correctly? Or is this a bug in the removal?

To Reproduce Code: https://gist.github.com/vitbaq/35f52f30bdfb442eb7a7a892ddb1d5d3 Log: https://gist.github.com/vitbaq/88a6f00edcadbb8724218a17551d113f

Testing environment:

• Operating System/Platform: Ubuntu 18.04.5 LTS (Bionic Beaver)
• Go version: go1.13.5 linux/amd64
• Grule-rule-engine version: 1.9.0

Hi, I'm a bit confused or lost with this scenario, wonder if you could guide me: Imagine I have 20 rules. And with a given fact, 3 of them pass. The conflict set chooses the highest priority out of those 3. However I need all 3 to be executed (to run the action or then clause).

Use case: I have a triggers/automations system that one trigger (an event like page created) could dispatch multiple actions (if the rules or conditions are met). Suppose that the user have created 2 automations (each automation consists of a trigger + GRL that has the action as a fact method) that their associated rules both pass for a given page creation event/trigger. One trigger/automation will send an email (one action or fact method), another will send a notification (another).

Couldn't find a way to achieve this. Is there a better way to do this or am I missing something?

Describe the bug Retracted rules are not restored when the engine is executed again. From the Retract method's documentation, I'm assuming that when we call the engine.Execute(dataCtx, knowledgeBase) the retracted rule should be restored. However, this is not happening.

To Reproduce Steps to reproduce the behavior: 1.Run the following code

package examples

import (
	"fmt"
	"strconv"
	"testing"

	"github.com/hyperjumptech/grule-rule-engine/ast"
	"github.com/hyperjumptech/grule-rule-engine/builder"
	"github.com/hyperjumptech/grule-rule-engine/engine"
	"github.com/hyperjumptech/grule-rule-engine/pkg"
	"github.com/stretchr/testify/assert"
)

type ComputedInputs struct {
	Name string
	Bid  float64
	Rate float64
}

const Grl = `
	rule SampleRule "Hero/Honda" {
		When
			ComputedInputs.Name == "Hero" || ComputedInputs.Name == "Honda"
		Then
			Log("SampleRule");
			ComputedInputs.Bid = ComputedInputs.Rate;
			Retract("SampleRule");
	}
`

func TestItems(t *testing.T) {
	testData := []*struct {
		CI      *ComputedInputs
		WantBid float64
	}{
		{
			CI: &ComputedInputs{
				Name: "Hero",
				Rate: 10,
			},
			WantBid: 10,
		},
		{
			CI: &ComputedInputs{
				Name: "Honda",
				Rate: 10,
			},
			WantBid: 10,
		},
	}

	// Prepare knowledgebase library and load it with our rule.
	lib := ast.NewKnowledgeLibrary()
	rb := builder.NewRuleBuilder(lib)
	byteArr := pkg.NewBytesResource([]byte(Grl))
	err := rb.BuildRuleFromResource("Tutorial", "0.0.1", byteArr)
	assert.NoError(t, err)

	engine := &engine.GruleEngine{
		MaxCycle: 10,
	}
	knowledgeBase := lib.NewKnowledgeBaseInstance("Tutorial", "0.0.1")

	for i, td := range testData {
		t.Run(strconv.Itoa(i), func(t *testing.T) {
			dataCtx := ast.NewDataContext()
			err := dataCtx.Add("ComputedInputs", td.CI)
			assert.NoError(t, err)

			rules, _ := engine.FetchMatchingRules(dataCtx, knowledgeBase)
			fmt.Printf("\nNo. of matching rules: %d\n", len(rules))

			knowledgeBase := lib.NewKnowledgeBaseInstance("Tutorial", "0.0.1")
			err = engine.Execute(dataCtx, knowledgeBase)
			assert.NoError(t, err)
			assert.Equal(t, td.WantBid, td.CI.Bid)
		})
	}
}

Expected behavior All the tests should pass

Additional context The tests are passing if we create a new instance of knowledgeBase on every execution. Do we have to create a new instance of knowledge base whenever a rule is retracted?

Hi there!

Is your feature request related to a problem? Please describe. With the release of 1.6.0, we saw the ability to use the same KnowledgeBase for multiple concurrent threads. This was great, since it allowed us to only parse the rule library once, which was a CPU-intensive operation that was bogging down our system previously. After upgrading grule we saw CPU usage reduce significantly. I think it can go further, though.

In my opinion, there are a couple of issues with the current implementation - namely KnowledgeBase cloning:

1. cloning still takes CPU cycles, and for our system, which sees potentially millions of events per day, this can add up to quite a bit of CPU usage.
2. cloning takes up quite a bit of memory, and may actually have a memory leak, unless the leak was actually caused by our own code, which is also possible (I didn't profile deep enough to make a strong claim). The point is, cloning takes a non-zero amount of memory per event, and at millions of events per day, the memory allocations are significant.

As far as I understand grule, I don't think cloning KnowledgeBases is necessary at all. A KnowledgeBase can be created and parsed once, say at the time of bootstrapping an application, and then traversed and evaluated as read only. This would require the evaluations - the results of the traversal - to be stored separately from the rule definitions in the KnowledgeBase.

Describe the solution you'd like Looking at the current code, I see the value of a rule being stored within that rule's type. This ties the rule instance to the evaluation/result instance, and since these are tied, the rule instance must be instantiated (cloned) per evaluation (hence KnowledgeBase.Clone()).

For example, here's the ExpressionAtom type, which contains the definition of a rule (or part of a rule), as well as the evaluation result:

// ExpressionAtom AST node graph
type ExpressionAtom struct {
	AstID   string
	GrlText string

	Variable     *Variable
	FunctionCall *FunctionCall
	Value        reflect.Value            // <-- evaluation result

	Evaluated bool                        // <-- evaluation result
}

If the evaluation results were kept in a different type, then that type could be instantiated per evaluation. This would be

• a lot cheaper memory-wise, since rule definitions aren't being copied in memory per execution, and
• it would save any CPU usage that's lost to cloning a KnowledgeBase.

A couple of ways to maintain results in a separate structure:

Option 1 Utilize the unique AstID on each rule definition type, and maintain a map of evaluation results that can be instantiated each time a KnowledgeBase is evaluated, e.g.

type evalResults struct {
	Evaluated       bool
	Value reflect.Value
}

// keep some map[string]*evalResults, and when a rule or partial rule is evaluated, write to this map using AstID as the key

One concern here is that this solution may require a sync.Map to enable concurrent writes of evaluation results, which could be less performant.

Option 2 More of an alternative approach to writes in Option 1, but a chan *evalResult could be used in place of a sync.Map, like this

type evalResults struct {
        AstID    string    // <-- add the AstID and pass it along on the channel
	evaluated       bool
	evalValueResult reflect.Value
	evalErrorResult error
}

// when rule is evaluated
evalResultsChannel <- &evalResult{ AstID: myAstID, ... }

// in some result collection routine
for _, evalResult := range evalResultsChannel {
    resultsMap[evalRules.AstID] = evalResult
}

This avoids any slowdown from a sync.Map, but result writes become async, so it's hard to know when a rule's evaluation result is formally stored in the resultsMap.

These are just a couple of ideas off the top of my head; there may be better solutions!

Is this decoupling of rule definition and evaluation result possible? Thanks in advance!

Describe alternatives you've considered none

Additional context none

Describe the bug I just upgraded from 1.5.0 to the latest 1.6.0 and started getting Max cycle reached, none of my rules had Retract. I'm also using JSON to build the rules, and it works as expected if I add Retract to the rules.

To Reproduce Steps to reproduce the behavior:

1. Use v1.6.0
2. Add one or multiple rule(s) without having Retract at the end for that/those rule(s)
3. See error Max cycle reached

Expected behavior Previous versions Retract wasn't mandatory

UPDATE: also getting warning while adding rule entry : <rulename>. got rule entry <rulename> already exist, possibly already added by antlr listener which wasn't happening before the upgrade UPDATE 2: It's also running the THEN statements for all rules that passed instead of just one (highest priority), kinda like what i was looking for here

Description:

• Current GruleV3 lexer and parser are generated using Antlt4 4.8, so upgraded to Antlr4 4.9.2
• There was a go runtime performance issue has been fixed in Anltr4 Library. so I have upgraded Antlr4 library to the recent commit

Reference: https://github.com/antlr/antlr4/pull/3243

The core problem in Golang runtime: it wastes too much time on hash code computing because of the incorrect Murmur3 and Set and hasher implementation. Compare the Java and the original Go runtime, Go computes murmur3 ~80,000,000 times, but the Java runtime only ~600,000 times.

Here's my benchmark in my MBP '16 2019, the g4 files are copied from https://github.com/antlr/grammars-v4/tree/master/sql/mysql/Positive-Technologies, the original Go runtime cost 4.8s, and this PR costs 171ms.

Changes to go.mod:

	github.com/antlr/antlr4/runtime/Go/antlr v0.0.0-20210928164016-886d244c73ef

Is your feature request related to a problem? Please describe. Currently my application is using grule-rule-engine and its working perfectly fine. Thanks @newm4n this is a great work !!!! Coming to the implementation, i am saving all my rule DSL's in DB (postgres) and pulling them when the container boot up and create a new knowledgeBase and keep it in Memory. so far all fine. let's suppose if someone creates a new rule or rules then i need to pull all the rules again and recreate the knowledgeBase and this is a heavy process even though i know that there are no changes to the existing knowledgeBase except new rules i have to add to the existing knowledgeBase.

Describe the solution you'd like What i am proposing is, when a new rule or rules are created i should be having a way to update the existing knowledgeBase with new rule instead of reloading all the N rules instead of creating a new knowledgeBase instance everytime with all the rules.

I can contribute on this, just that i need some ideas or pointers from you to implement this feature :)

Additional context Let me know if you need more info

I think it should be easy to support logic NOT (!) operator.

The rule would looks like:

rule RuleOne "Some rule description." salience 10 {
    when
        !(someContext.attributeA.Match("123") &&
          someContext.attributeA.Contains("1"))
    then
        ...

Without NOT operator, I have to write the rule like:

rule RuleOne "Some rule description." salience 10 {
    when
        someContext.attributeA.Match("123")==false ||
        someContext.attributeA.Contains("1")==false
    then
        ...

I try to create two simple rules similar with the tutorial stated in the docs. These are my rules:

// Initial fact
// MF.IntAttr = 1200000

rule Rule1 "Desc of Rule1" salience 0 {
	when
		MF.IntAttr >= 100000
	then
		Log("Rule1 called");
		Retract("Rule1");
}

rule Rule2 "Desc of Rule2" salience -1 {
	when
		MF.IntAttr / 500000 > 0.0
	then
		Log("Rule2 called");
		MF.IntAttr = MF.IntAttr - 500000;
}

I expect the engine to re-evaluate the condition in the Rule2 whenever a new cycle begins. And at some point, the condition will be false and the cycle done. But instead, I got an error like this:

panic: GruleEngine successfully selected rule candidate for execution after 5000 cycles, this could possibly caused by rule entry(s) that keep added into execution pool but when executed it does not change any data in context. Please evaluate your rule entries "When" and "Then" scope. You can adjust the maximum cycle using GruleEngine.MaxCycle variable.

It seems the engine is not re-evaluating the condition in Rule2 and remember it as true causing the cycle to keep repeating again and again. Please advice if I am doing it wrong. Thanks.

Describe the bug when using a map in the data context and attempting to reference it in a rule GRL errors occur. For example the rule:

rule testRule "set internet revenue" salience 10 {
    when
        ORDER["IntAttribute"] == 123
    then
        OUTPUT["result"] = "testResult";
        Retract("testRule");
}

Fails with the following errors:

line 4:9 token recognition error at: '['
line 4:24 token recognition error at: ']'
line 4:10 mismatched input '"IntAttribute"' expecting THEN
line 6:10 token recognition error at: '['
line 6:19 token recognition error at: ']'
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '"IntAttribute"'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '=='  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '123'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected 'then'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected 'OUTPUT'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '"result"'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '='  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '"testResult"'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected ';'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected 'Retract'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '('  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected '"testRule"'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected ')'  lib=grule struct=GruleParserV2Listener
ERRO[0000] GRL error, after '[salience 10 { when ORDER]' and then unexpected ';'  lib=grule struct=GruleParserV2Listener
ERRO[0000] Loading rule resource : Byte array resources 170 bytes failed. Got GRL error, after '[salience 10 { when ORDER]' and then unexpected ';'. Time take 2 ms 
panic: error were found before builder bailing out. Got GRL error, after '[salience 10 { when ORDER]' and then unexpected ';'

To Reproduce Steps to reproduce the behavior:

1. Use the following code:

package main

import (
	json "encoding/json"
	"fmt"

	ast "github.com/hyperjumptech/grule-rule-engine/ast"
	builder "github.com/hyperjumptech/grule-rule-engine/builder"
	engine "github.com/hyperjumptech/grule-rule-engine/engine"
	pkg "github.com/hyperjumptech/grule-rule-engine/pkg"
)

func main() {
	// Create example JSON
	inputjson := `{
		"IntAttribute": 123,
    	"StringAttribute": "Some string value",
    	"BooleanAttribute": true,
    	"FloatAttribute": 1.234,
    	"TimeAttribute": "2020-03-11T10:00:00Z",
	}`

	// unmarshal the json into a map
	var input map[string]interface{}
	json.Unmarshal([]byte(inputjson), &input)

	// create the new Data Context
	dataCtx := ast.NewDataContext()
	// Add the input
	err := dataCtx.Add("INPUT", &input)
	if err != nil {
		panic(err)
	}
	// add an output map
	var output map[string]interface{}
	err = dataCtx.Add("OUTPUT", &output)
	if err != nil {
		panic(err)
	}

	workingMemory := ast.NewWorkingMemory()
	knowledgeBase := ast.NewKnowledgeBase("test", "1.0.0")
	ruleBuilder := builder.NewRuleBuilder(knowledgeBase, workingMemory)

	// example rules
	drls := `
		rule testRule "set internet revenue" salience 10 {
			when
				ORDER["IntAttribute"] == 123
			then
				OUTPUT["result"] = "testResult";
				Retract("testRule");
		}
	`

	byteArr := pkg.NewBytesResource([]byte(drls))
	err = ruleBuilder.BuildRuleFromResource(byteArr)
	if err != nil {
		panic(err)
	}

	engine := engine.NewGruleEngine()
	err = engine.Execute(dataCtx, knowledgeBase, workingMemory)
	if err != nil {
		panic(err)
	}

	fmt.Println(output["result"])
}

2. compile and run.

Expected behavior Expected the rule to run and produce the output testResult

How to apply default value if no rule matched ? add a default rule like ?

rule is_girl "set girl" salience 0 { when mf.sex == "famale" then mf.result = "girl" }

rule is_boy "set boy" salience 0 { when mf.sex == "male" then mf.result = "boy" }

rule default "set default value if no rule matched" salience 0 { when true then mf.result = "unknown"; }

Thanks

Describe the bug I want to print the rulename which got successfully executed based on the given input Fact. I tried to use ruleEntries, err := engine.FetchMatchingRules(dataCtx, knowledgeBase) But ruleEntries returns an empty array

To Reproduce Steps to reproduce the behavior: This is my code for Evaulation

func Evaluate(stock *StockInfo) string {

	dataCtx := ast.NewDataContext()
	err := dataCtx.Add("stock", stock)
	if err != nil {
		panic(err)
	}
	knowledgeLibrary := ast.NewKnowledgeLibrary()
	ruleBuilder := builder.NewRuleBuilder(knowledgeLibrary)
	fileRes := pkg.NewFileResource("StocksEvaluation/stock-rule.grl")
	err = ruleBuilder.BuildRuleFromResource("StockEvaluator", "0.0.1", fileRes)
	if err != nil {
		panic(err)
	}
	knowledgeBase := knowledgeLibrary.NewKnowledgeBaseInstance("StockEvaluator", "0.0.1")
	engine := engine.NewGruleEngine()

	err = engine.Execute(dataCtx, knowledgeBase)
	if err != nil {
		panic(err)
	}
	ruleEntries, err := engine.FetchMatchingRules(dataCtx, knowledgeBase)
	if err != nil {
		panic(err)
	}
	for _, ruleEntry := range ruleEntries {
		println(ruleEntry.RuleName)
	}
	return stock.Decision
}

2. My Fact Object

type StockInfo struct {
	StockName       string
	CashFlow        float64
	Profit          float64
	MarketCap       float64
	EnterpriseValue float64
	Dividend        float64
	PromoterHolding float64
	Decision        string
}

func (stock *StockInfo) ShouldBuy(yesorno bool) string {
	//print(yesorno)
	if yesorno == true {
		return fmt.Sprintf("Buy %s", stock.StockName)
	} else {
		return fmt.Sprintf("Don't Buy %s", stock.StockName)
	}

}

My grl Rules

rule CashFlowVsNetProfit "Cash Flow is greater than Net Profit" salience 10 {
    when
        stock.CashFlow >= stock.Profit
    then
        Log("CashFlowVsNetProfit evaluated");
        stock.Decision = stock.ShouldBuy(true);
        Retract("CashFlowVsNetProfit");
}
rule MarketCapVsEnterpriseValue "If the enterprise value is way more than market cap, then it can be an alarming sign." salience 8 {
    when
        stock.EnterpriseValue > stock.MarketCap
    then
        Log("MarketCapVsEnterpriseValue evaluated");
        stock.Decision = stock.ShouldBuy(false);
        Retract("MarketCapVsEnterpriseValue");
}

rule DividendYield "a company that gives a higher dividend might imply that it is not utilising its profits for growth." salience 7 {
    when
        stock.Dividend >= 3.0
    then
        Log("Dividend Yield is evaluated");
        stock.Decision = stock.ShouldBuy(false);
        Retract("DividendYield");
}

I see that the CasFlowVsNetProfit gets successfully evaluated based on my fact input as the log message gets printed. But the array is still an empty object

Expected behavior The array should be an empty object and it should print the rule evaluated.

Additional context None

I notice there is a similar Issue which was closed. However, I think there is some use case which can still be consider supporting interface / pointer.

I'm currently providing the Facts through REST API which can be in any arbitrary form, hence map[string]interface{}. But rule evaluation gives me

can not use data type of interface in EQ comparison

or

can not use data type of ptr in EQ comparison

if I use pointer.

Solution I propose that in reflectmath.go/Evaluate* handles these "reference" type with e.g.:

func EvaluateEqual(left, right reflect.Value) (reflect.Value, error) {
	if left.Kind() == reflect.Pointer || left.Kind() == reflect.Interface {
		return EvaluateEqual(left.Elem(), right)
	}

	if right.Kind() == reflect.Pointer || right.Kind() == reflect.Interface {
		return EvaluateEqual(left, right.Elem())
	}
....
}

Would gladly make PR contribution but I would like to open this for discussion first to know what are the implications and if it's aligned with the design.

Describe the bug Not Bug

Hi，noticed the latest editor has merged into master last week. After run Main.go in /editor/cmd, have no idea how to play in the playgroud.

Hope anyone can share useful demos. Thx

Is your feature request related to a problem? Please describe. Noticed that I can visit a JSON Fact as a normal object or a map, but I can't visit a map like a normal object.

Describe the solution you'd like

    drls := `rule Demo "demo" salience 10 {
    when 
        myMap.name == "foo"
    then
        myMap.name == "bar";
	Retract("Demo");
    }`

    myMap := map[string]string{
        "name": "foo",
    }
    err := dataContext.AddMap("myMap", myMap)

Due to the restrictions of the Go, there is no way to set the value of a struct's non-exported filed. If this feature can be supported, I am able to use non-exported fields when defining my DSL.

Describe alternatives you've considered An alternative solution is define a tag, for example:

type User struct {
    Name string `grule:"name"`
}

    drls := `rule Demo "demo" salience 10 {
    when 
        user.name == "foo"
    then
        user.name == "bar";
	Retract("Demo");
    }`

    user := User{
        Name: "foo",
    }
    err := dataContext.Add("user", &user)

If the tag is specified, the grule engine can use it When calculating the rules.

Additional context

type User struct {
    Name string `json:"name"`
}

{
    "name": "foo"
}

Most APIs are defined in snake case. So I wish can define the rules in snake case.

Describe the bug

Loading rules using pkg.NewFileResourceBundle() results in:

panic: runtime error: invalid memory address or nil pointer dereference
[signal 0xc0000005 code=0x0 addr=0x40 pc=0x627659]

goroutine 1 [running]:
github.com/hyperjumptech/grule-rule-engine/engine.(*GruleEngine).ExecuteWithContext(0xc00011de20, {0x801a68, 0xc000022088}, {0x805fb0?, 0xc00015ec00}, 0x0?)   
        C:/Users/tashi/go/pkg/mod/github.com/hyperjumptech/[email protected]/engine/GruleEngine.go:86 +0x59
github.com/hyperjumptech/grule-rule-engine/engine.(*GruleEngine).Execute(...)
        C:/Users/tashi/go/pkg/mod/github.com/hyperjumptech/[email protected]/engine/GruleEngine.go:52
main.main()
        C:/Users/tashi/go/src/gitlab.com/raaho/fin-service/main.go:47 +0x2f0

To Reproduce

Steps to reproduce the behavior:

1. I have copied the tutorial code and tried to load the rules from an external .grl file. I used pkg.NewFileResourceBundle() to load it.
2. Instead of running without errors and printing the updated fact, I see the given error.
3. When I give the absolute path of rules, the code works without errors.

Expected behavior

The code compiles without errors and displays updated fact.

Additional context My directory:

- main.go
- rules/
  - test.grl

Describe the bug Am using JSON Fact and appending values to an array. The array is not getting updated in the datacontext

JSON Fact:

{
	"status": "start",
	"payloadData": "Whatever data",
	"logs": ["Hello"]
}

Rule:

rule Testing {
		when 
			data.status == "start"
		then 
			data.status = "stop";
			data.logs.Append("PERFECT1");
	}

Expected:

{
	"auditlogs": ["Hello", "PERFECT1"],
	"payloadData": "Whatever data",
	"status": "stop"
}

Actual Result:

{
	"auditlogs": ["Hello"],
	"payloadData": "Whatever data",
	"status": "stop"
}

To Reproduce Steps to reproduce the behavior:

package main

import (
	"encoding/json"
	"fmt"

	"github.com/hyperjumptech/grule-rule-engine/ast"
	"github.com/hyperjumptech/grule-rule-engine/builder"
	"github.com/hyperjumptech/grule-rule-engine/engine"
	"github.com/hyperjumptech/grule-rule-engine/pkg"
)

func main() {
	dataContext := ast.NewDataContext()
	err := dataContext.AddJSON("data", []byte(`{"status" : "start", "payloadData" : "Whatever data", "logs":["Hello"]}`))
	if err != nil {
		fmt.Println(err)
	}

	rule := `
	rule Testing {
		when 
			data.status == "start"
		then 
			data.status = "stop";
			data.logs.Append("PERFECT1");
	}`

	lib := ast.NewKnowledgeLibrary()
	rb := builder.NewRuleBuilder(lib)
	rb.BuildRuleFromResource("TestJSONBitComplex", "0.0.1", pkg.NewBytesResource([]byte(rule)))
	kb := lib.NewKnowledgeBaseInstance("TestJSONBitComplex", "0.0.1")
	eng := &engine.GruleEngine{MaxCycle: 5}
	err = eng.Execute(dataContext, kb)
	if err != nil {
		fmt.Println(err)
	}

	result := dataContext.Get("data").Value().Interface().(map[string]interface{})
	b, err := json.Marshal(result)
	if err != nil {
		fmt.Println(err)
	}
	fmt.Println(string(b))
}

Expected behavior The result printed should have the appended values in the logs field