Research data
GNUplot index/every selection map
Map source lines to dataset, block and point positions selected by index/every.
Your inputs
Start with an exampleYour result
Example resultHow it works
Ignore leading blank records. One blank separates blocks; two blanks separate datasets, even with comments between them. Number data records from zero inside each block, including unplottable records. Test the literal index, block and point predicates and report rejected conditions with physical source lines.
The example has two datasets and two blocks in each. every 2:1:1 selects point 1, then every second point: source lines 3, 7, 11 and 15. The record bad still consumes its ordinary point position.
Limits
Ordinary non-VMS external text, default whitespace fields and # comments only. Up to 1 MiB UTF-8, 50000 physical records, 10000 characters per record and an 8 MiB complete report. LF/CRLF only; no bare CR, NUL or other ASCII controls except TAB. Empty and comment-only files are valid. Canonical nonnegative decimal selectors are limited to 2147483647; increments must be positive. index accepts m, m:n, m:n:p and m::p. every accepts up to six fields with omitted interior defaults; no trailing colon. Logical source point ordinals also cover rejected blocks. All input is scanned, including datasets beyond a bounded index. No column parsing, plotting, headers, skip, custom separators/comments, expressions, negative aliases, named indexes, binary/matrix, inline input or datablocks. A source map does not establish plot success or reconstruct an equivalent filtered data file.