Dev Updates
Development in the open. PyneCore and PyneComp releases land here as they happen — what got fixed, what got added, and now and then what we got wrong — along with progress on work that is not finished yet. Synced directly from our Discord dev-updates channel.
PyneCore’s full changelog lives on GitHub; PyneComp’s progress you will find here and nowhere else. More granular and more technical than News — this is the engineering log, not the announcements.
PyneComp v6.0.82 — scripts that TradingView rejects are now refused with a clear error,…
import someone/class/1, an input.source whose default is another input, and an undeclared type such as array.new<foo> now stop with a compile error at the right line instead of producing a script that fails when it runs…
Read more →PyneComp v6.0.81 — Compiled and converted scripts now match TradingView more closely on…
A var or varip variable initialized with an if, switch, for or while block now runs that block only until it yields a value, then never again, as on TradingView - A v6 for loop whose to bound calls a function now…
Read more →PyneCore v6.13.2 released — Pine semantics measured on TradingView: exact constant…
bool() of a constant is exact (bool(1e-10) is true), and v4/v5 bool(na) stays na (#107) - for loops: a by of zero or below is an error, also on descending loops; a fractional counter's end is compared with Pine's…
Read more →PyneCore v6.13.1 released — coin-flip backtests now run in parallel and stop early once…
pyne robustness --jobs runs the per-draw coin-flip backtests in parallel processes; the results do not depend on the worker count - --early-stop edge|better|off ends a coin-flip test as soon as its verdict is clear…
Read more →PyneCore v6.13.0 released: pyne robustness tests how much of a strategy's backtest could…
pyne robustness and the new pynecore.analysis package: trades measured in R from their initial stop, bootstrap and Monte Carlo with ruin and target odds (optional --max-leverage), extra-cost sensitivity, calendar…
Read more →PyneComp v6.0.80: more published Pine scripts now convert and compile cleanly, and inputs…
A global x := input.() or x += input.() always ran with its default value and ignored your settings. It now reads your setting, and x keeps its earlier value until the reassignment. - Scripts with an unused helper…
Read more →PyneCore v6.12.1 released — Sharpe and Sortino ratios now match TradingView, display. and…
The documentation is part of the installed package; pynecore.docs_path() returns the docs of the installed version - display.pine_screener is available
Read more →PyneCore v6.12.0 released — the transform pipeline now runs in Rust, and math.sum /…
Every AST transform step has a native Rust port that produces the same tree as the Python step, shipped in the platform wheels - PYNE_AST_BACKEND selects native (default), python or verify; without the extension…
Read more →PyneComp v6.0.79 — scripts with type or argument mistakes now get the same compile errors…
Mixing incompatible types in an expression ("a" * 2, "a" > "b", not 1.5), a non-bool if/while condition, or a typed declaration with the wrong literal (float f = "a") is now reported as a compile error instead of being…
Read more →PyneCore v6.11.1 released — fixes for scripts sharing a process with other scripts'…
A script now runs only the libraries it imports; earlier scripts' libraries in the same process no longer run on every bar or merge their request.security contexts (#98) - Daily request.security values on 24/7 symbols…
Read more →