Module quality_trends 

Trends and gaps: a report’s segments in order, pooled into the bars a terminal has room for, and the time windows a study said to expect checked against the segments a run counted. Built from the measurements a run kept: nothing here reads.

Structs

BarChange

How a bar’s rate stands against another’s.

GapCheck

The stated windows, checked against the segments a run counted.

GapRun

Consecutive expected windows that share a kind of gap.

TrendBar

One bar: the consecutive segments it pools, and how much of them the run saw.

TrendRow

One line of the Trends table: the rows each segment holds, the rows the sample drew, or a column’s measure, pooled into bars of consecutive segments.

TrendSlot

One segment in the order Trends draws them: one a run profiled, or one it counted rows in and sampled none of.

TrendView

The Trends table for a measure, as wide as it is drawn.

Enums

GapKind

Why an expected window has no rows to show.

Gaps

What checking the stated windows came to.

TrendMeasure

What a Trends line measures.

Constants

MAX_EXPECTED_WINDOWS

The most windows a stated range is checked over. A daily grain over half a century, or an hourly one over two years; past it, a coarser grain or a shorter range.

MAX_GAP_RUNS

The most runs of gap windows listed; the rest are counted.

Functions

bar_change

line’s value in bar beside its value in other, judged as Segments judges a change: a point or more, past sampling noise unless the counts are exact. A rows line moves by its count, so only rates are judged.

bar_span

What a bar spans: the calendar range of its windows on a time grain, and its first and last segment otherwise. Rows with no time, which sort last, are said apart from the calendar.

calendar_span

A span of windows in calendar terms, inclusive: 2024-01-01 to 2024-01-28 for days, weeks and months, to the minute for hours.

compared_bar

The bar a bar is compared with: the one holding the baseline segment when the plan compares with one, and otherwise the bar before.

expected_gaps

The expected windows of plan checked against results, or None when the plan states none. Windows with rows are found among the segments; one the sample missed is told from one with no rows by the exact count the run took, and one outside the time range the scope reads is out of scope, never empty.

floor_window

The start of the window time falls in, cut as a run cuts them: hours on the hour, days at midnight, weeks from Monday, months from the first.

next_window

The start of the window after the one starting at start.

segment_coverage

Whether the segments are a sample’s, then the segments with rows the sample drew none of and the segments it drew few of: what TrendView::coverage counts, without building the view.

trend_slots

Every segment in order, the ones the sample missed among the ones it drew.

trend_view

The Trends table for metric, bars wide: rows per segment first (the exact count, then on a sample the rows it drew), then every column the measure is above zero in somewhere, the one that moves most first. Each bar pools consecutive segments (their counts over their rows), so a daily grain over years reads as years, and a thin day’s sample does not make a bar alone. A segment the sample missed is a slot like any other, so a bar of them says so rather than vanish.

wilson_interval

The 95% Wilson score interval for count of n: where the whole’s rate likely sits, given a simple random sample. Unlike the normal interval it stays inside 0 to 1 and does not shrink to nothing at a count of zero, which a thin sample often has.

window_start

Where window label starts, read back from the label a run gave it: the inverse of time_window_label. None for the rows with no time, and for a label that is not a window’s.