bug:invalid jump position:0 about tengo HOT 13 OPEN

Too large an array. See related issue: #402

from tengo.

OK，thx

from tengo.

This may not be the same problem as that one，I changed my code to the following way of writing and still reported the same error：
for i in range(0, 9) { //fmt.Println(list[i]) }
I traced the source code and found that the problem lies in the MakeInstruction() of 'the instructions. go' file.
In line 28, n:=uint16 (o), the actual value of o is 66114, which exceeds the maximum value of uint16 (65535).
Is there any way I can correct it？

from tengo.

aha, can you try a larger uint and see if that helps?

from tengo.

I changed the 4 Jump instructions from uint16 to uint32, and changed the 2 bytes to 4 bytes. They are now working properly

And thank you again for your reply

from tengo.

However, the execution efficiency is slightly lower than before. I used the for loop to execute i++ 100 million times, which is about 200 milliseconds slower than before.

from tengo.

Do you have a measure of the relative slowdown?

from tengo.

func main() {
src := `
fmt:=import("fmt")
V_INTERNAL_F := func() {
index:=0;
for i:=0;i<100000000;i++{
index++
}

return index

}
V_RETURN := V_INTERNAL_F()

`
start := time.Now()
res, err := run(src)
if err != nil {
panic(err)
}
end := time.Since(start)
fmt.Println("cost:", end)
//println(res)
j_str, err := json.Marshal(res)
if err != nil {
panic(err)
}
fmt.Println(string(j_str))
}

from tengo.

I executed the above code 20 times using the pre modified code and the modified code, and took an average value for comparison

from tengo.

Sorry, to clarify, 200ms is a lot if your baseline was say 500ms, but negligible if it was say 10 seconds. What does the 200ms compare against?

from tengo.

The baseline is 7 to 8 seconds, and 200 milliseconds is the difference in code execution time before and after modification, shouldn't it be too much?

from tengo.

That’s a 2-3% difference. If it’s consistently repeatable we should try to improve it if possible but if not, the trade-off might be worth it

from tengo.

I submitted a pull request for this: #433

from tengo.