Known issue in PineForge Codegen 1.9.1: str.tostring of built-in boolean flags, and of variables holding them or a ta.* boolean result, returns "1" / "0" #315

In Pine Script, str.tostring of a boolean returns "true" or "false". In PineForge 1.9.1, some boolean values print as "1" or "0" instead. In every case we measured, the boolean values themselves are correct; only their text is wrong. Any string built from that text is affected, for example an order comment.

Measured as wrong in 1.9.1 (the list is a lower bound):

• the built-in flags: str.tostring(session.isfirstbar), session.islastbar, session.isfirstbar_regular, session.islastbar_regular, barstate.islast, barstate.isfirst, barstate.isnew, barstate.isconfirmed, barstate.islastconfirmedhistory, barstate.isrealtime
• a variable that holds one of these flags, for example fl = session.isfirstbar followed by str.tostring(fl)
• a variable that holds a boolean ta.* result, for example ok = ta.crossover(close, ta.sma(close, 5)) followed by str.tostring(ok). The same happens with ta.crossunder, ta.cross, ta.rising and ta.falling, and with a variable assigned from a user function that returns such a result.
• concatenation, for example "first=" + str.tostring(session.isfirstbar) returns first=0

Measured as correct in 1.9.1:

• str.tostring(true), str.tostring(false), a var bool
• a comparison, inline or stored (x = close > open), a logical expression, a boolean carried across bars (including x[1]), and a boolean assigned in an if
• a user function returning a comparison, a boolean field of a user-defined type, an array<bool> element and a map<string, bool> value
• fl = barstate.islast and close > open
• ta.crossover, ta.crossunder, ta.rising and ta.falling used directly inside str.tostring, for example str.tostring(ta.crossover(close, ta.sma(close, 5)))
• str.format with boolean arguments, for example str.format("{0} {1}", close > open, session.isfirstbar_regular) prints true / false for both arguments

Example, measured on NYSE:F 60-minute bars from 2024-01-02 to 2024-02-29: a script writes str.tostring(session.isfirstbar_regular) and str.tostring(session.islastbar_regular) into 97 exit comments. TradingView printed false / true. PineForge 1.9.1 printed 0 / 1, so the text matched in 0 of 97 comments while the values agreed in all 97.

Not yet checked: other boolean ta.* calls (for example the pivot functions), boolean members of syminfo.* and timeframe.*, request.* results, and these values stored in a user-defined type field or an array.

Workaround: none verified yet. A fix is in progress.

Addition 2026-11-03: further cases measured in 1.9.1.

Also wrong ("1" / "0"):

• a function parameter that receives session.isfirstbar or a boolean ta.* result, passed to str.tostring inside the function (a parameter that receives barstate.isconfirmed printed correctly);
• an array<bool> element created with array.from(...) from such a flag or ta.* result;
• a variable holding a boolean request.security result.

Wrong in a different way: a variable holding timeframe.isintraday printed a non-printable character (byte 0x01) instead of true. Variables holding timeframe.isdaily, timeframe.isweekly, timeframe.ismonthly or timeframe.isseconds printed an empty string instead of false. These were measured on a 1-hour chart.

Measured as correct: a boolean field of a user-defined type and a map<string, bool> value assigned from a flag or a ta.* result; a stored not na(ta.pivothigh(...)); a stored ternary; a parameter that receives a comparison; and timeframe.isintraday used directly inside str.tostring.