cpl run
check, then run in the debug interpreter, reading the input file or standard input; a file with an `@interactor` runs against it, on the hidden test (`cpl run --help` has the options for traces); `--release` runs the C++ that `cpl build` writes
usage: cpl run <file.cp> [input] [options]
Checks the file, then runs it in the debug interpreter on the input file
or on standard input. What `debug(...)` shows goes to standard error.
options:
--trace [SPEC] record the run: every change to a top-level variable,
every print and every debug(...); with SPEC, comma-
separated fn names or globs (`*`, `?`) with `!pat`
exclusions (e.g. `solve,!helper`). Writes
<file>.trace.html unless --format says where it goes.
With no input named and none piped in, it records the
case `cpl stress` saved last for the file
--test SPEC with --trace: record only these 1-based multitest cases
(e.g. `3`, `1,3`, `2..=4`)
--watch VAR with --trace: record steps only when VAR changes (repeatable
or comma-separated, e.g. `--watch dp,ans`)
--last N with --trace: keep only the last N steps (ring buffer)
--depth N with --trace: record only up to call depth N (0 = top level)
--format json without --trace: the value of each top-level variable,
the output and the time, for an editor.
with --trace: the trace as JSON on standard output
--format html with --trace: the trace viewer with the trace in it,
one self-contained page on standard output
--live with --trace: serve the viewer on localhost and record
again each time the file is saved, on the same input;
with no input named, the case `cpl stress` saved last.
The page keeps its place and shows the first change
--brute run the file's `@brute(main)` fn in place of the solution
--release run what a judge runs: the C++ of `cpl build`, built with
the judge's flags (those g++ here accepts) on the judge's
stack, and kept until the file changes.
Fast, and without the interpreter's checksGenerated from cpl 0.2.2 (dev, a0d0078e).