# Exit codes Every `samplekit` command ends with one of these codes. Messages go to the error stream (stderr); the data a command was asked for goes alone to the output stream (stdout). | Code | Meaning | | --- | --- | | 0 | Success | | 1 | A usage error: what was asked does not exist or cannot be done | | 2 | A data error: something in the data is wrong | | 3 | The system refused to read or write a file, or the machine cannot run the model | | 130 | Interrupted with Ctrl-C | ## 0 · Success - A selection that matches nothing. - An export of an empty selection: its header is written. - A command with nothing to do, which says `nothing to change` and why: removing a tag no sample carries, renaming a tag to itself, a `restore` with nothing to take back, a `compute` with nothing to compute, a `set`, `tag` or `remove` whose filter selects no sample. - A command that printed warnings, such as, with `-v`, a file in the folder that is not a sample (a Markdown file without frontmatter). - A question answered no (`n`, an empty line, the end of the input), which says `nothing written`: nothing failed. A script that must stop there writes with `--write`, or reads what the command printed. ## 1 · Usage error - An unknown option, command, field, query, profile, export or figure. The message suggests the nearest name. - A filter that does not parse. - A name `new` cannot give a sample, or an existing sample `new` would overwrite. - `log --script` on a snapshot that no script made. - `--accept waiting`: a value waiting for an input fails nothing, so there is nothing to accept. - `remove` given a folder alone, or nothing: it deletes the files named, or those a filter selects. ## 2 · Data error - A file named on the command line is not a sample SampleKit can read. Over a folder, the samples that could be read are still shown, the others are named, and the code is 2. - The `.samplekitrc` that applies has an error (an unknown key, a misspelled section): it is set aside, said in a warning, the samples are read without it, and the code is 2. - A formula raised an error during `compute`. - A script of the model, run by `run`, raised an error, or asked with `input()` where nothing could answer. Nothing is written. - `validate` found a defect. - A figure has nothing to draw, such as no sample holding both axes. - A file changed on disk between the moment SampleKit read it and the moment it was about to write it: a sample, an export, the `.samplekitrc` that `init` writes. Nothing is written. - `status --exit-code` found a value that is not current. A value waiting for an input nobody entered does not count; with `--accept edited`, neither does a value typed over its formula, and with `--accept edited:area,young_modulus` those values alone. - A file was written from values that are not current, by `export --write`, a table's `-o FILE --write` or `plot -o FILE --write`: the file is written, and the code is 2 so that a script stops there. `--accept` followed by their states makes it 0: `outdated`, `edited`, `edited:NAME,…`, `failed`, `never_computed`, a comma between. - `export --write` over a selection where a sample could not be read: nothing is written, since its rows would be missing, and the file there is kept. ## 3 · The system refused The operating system refused a file or a folder: it does not exist, the permission to read or write it is missing, or a symbolic link cannot be followed. The message names the file. A file that exists but is not a valid sample is a data error, 2. The machine cannot run the model when its environment lacks what it needs: no Python where `[model] python` says (nothing else is searched for), the `samplekit` package missing from it, or another version of it than the command line's: the message names both versions. That is the machine's to mend, as a file refused is, not the data's. ## 130 · Interrupted Ctrl-C stops the command, at its question too. What a computation finished before is kept; the value in progress is not written.