PineScript升级至V4时出现未声明标识符错误

I see you're hitting a common snag when upgrading from PineScript V1 to V4: the undeclared identifier 'csf' error on line 21. This issue stems from a key behavior change between versions regarding persistent variables in custom functions. Let's break down the fix and explain what's going on.

The Root Cause #

In PineScript V1, when you referenced a variable's historical value (like csf[1]) inside a function, the compiler automatically treated that variable as persistent (it retained its value across bars). However, PineScript V4 requires explicit declaration of persistent variables using the var keyword within functions—without it, the variable is reinitialized on every bar, and csf[1] refers to a variable that never got declared.

Fixed Full Code #

Here's the upgraded script with all necessary changes to work in V4:

indicator(title="DecisionPoint Price Momentum Oscillator [LazyBear]", shorttitle="DPMO_LB")
src = input(close, title="Source")
length1 = input(35, title="First Smoothing")
length2 = input(20, title="Second Smoothing")
siglength = input(10, title="Signal Smoothing")
fr = input(true, title="Fill Region")
ehc = input(false, title="Enable Histo Color")
ebc = input(false, title="Enable Bar Colors")
soh = input(false, title="Show Only Histo")
slvl = input(false, title="Show OB/OS Levels")
oblvl = input(2.5, title="OB Level")
oslvl = input(-2.5, title="OS Level")

calc_csf(src, length) =>
    var float csf = na  // Explicitly declare persistent variable with var
    sm = 2.0 / length
    csf := (src - nz(csf[1])) * sm + nz(csf[1])
    csf

i = (src / nz(src[1], src)) * 100
pmol2 = calc_csf(i - 100, length1)
pmol = calc_csf(10 * pmol2, length2)
pmols = ta.ema(pmol, siglength)
d = pmol - pmols

duml = plot(not soh and fr ? (d > 0 ? pmol : pmols) : na, style=plot.style_circles, color=color.gray, linewidth=0, title="DummyL")
plot(0, title="MidLine")

hc = d > 0 ? (d > d[1] ? color.lime : color.green) : (d < d[1] ? color.red : color.orange)
plot(d, style=plot.style_columns, color=ehc ? hc : color.gray, title="Histo")

sigl = plot(soh ? na : pmols, title="PMO Signal", color=color.gray, linewidth=2)
mdl = plot(soh ? na : pmol, title="PMO", color=color.black, linewidth=2)

fill(duml, sigl, color.new(color.green, 70), title="PosFill")
fill(duml, mdl, color.new(color.red, 70), title="NegFill")

barcolor(ebc ? hc : na)

plot(not soh and slvl ? oblvl : na, title="OB Level", color=color.gray, linewidth=2)
plot(not soh and slvl ? oslvl : na, title="OS Level", color=color.gray, linewidth=2)

Key Modifications #

• Replace study() with indicator(): While study() still works in V4 as an alias, indicator() is the recommended modern syntax.
• Declare csf with var: In the calc_csf function, we added var float csf = na to mark the variable as persistent. This ensures it retains its value across bars, allowing csf[1] to reference the previous bar's value without an undeclared identifier error.
• Use := for variable assignment: V4 requires the := operator for reassigning values to persistent variables (instead of = which is for initial declaration).
• Prefix ema with ta.: In V4, built-in technical analysis functions are grouped under the ta namespace, so ema() becomes ta.ema().
• Clean up input syntax: Split the oblvl and oslvl inputs into separate lines for readability (the original comma-separated syntax was valid, but this is cleaner).
• Fix nested ternary formatting: Added parentheses to nested ternary conditions for better readability, which helps avoid unexpected behavior.

Additional Notes #

Always test the upgraded script against the original V1 version to ensure the calculations match exactly—this is a critical step to confirm the migration didn't break the indicator's logic.

内容的提问来源于stack exchange，提问作者Process PP