Kanta output is now visible without any application logging setup: the
event loggers Kanta owns (kanta.bootstrap/migration/transaction) are
set to INFO instead of inheriting the root level, which defaults to
WARNING and hid INFO transaction logs. The plain kanta diagnostics
logger stays unset so root controls its verbosity. Tests that relied
on INFO events being suppressed now scope their caplog assertions to
the relevant logger.
-r is now repeatable and restrictive: unit specs (l/s/v/r) intersect
their database line windows, while plain-number specs index into the
filtered output items left after unit windows and --grep, acting as a
pure display filter (negatives from the end, single index errors with
exit 2, ranges truncate). --context N (default 0) then expands each
surviving item GNU grep-style by N surrounding in-window records, free
of the grep/plain filters, merging overlaps.
Revisions are absolute state positions between diffs: the empty initial
state is revision 0 and each change record advances it by one. Snapshots
persist the revision of the state they contain in a new r field, stamped
with a meta format version k (kanta.structs.META_FORMAT, currently 1);
change records carry no revision of their own.
- kanta.rev exposes the current revision (raises before open)
- replay restores the revision, trusting r on k-marked snapshots and
counting change records from the start of the file for legacy (k0) ones
- rotation preserves absolute revisions, counted from record positions
- meta migration k1 rewrites a legacy file in full on writable open:
leading snapshot preserved and re-stamped, all change records kept,
closing snapshot with the total revision; read-only opens compute the
same revision without touching the file
- CLI range selectors accept rN bounds: rN is the record that produced
state N with its state (the change, or the snapshot materializing it);
single indexes must exist, ranges truncate Python-style
- CLI snapshot indicator lines show the snapshot revision
configure_logging() now runs with default arguments when kanta.logging
is imported, attaching a plain stderr handler with propagate=False to
the event loggers (kanta.bootstrap/migration/transaction) that carry
Kanta-rendered output. No levels are set by default, so event output
inherits the effective root level: a framework switching root between
INFO in development and WARNING in production governs Kanta output too.
Other configure_logging changes: channel enable flags use
logger.disabled (propagate toggling no longer silences now that event
loggers have their own handler), skiproot=False removes Kanta's
handler and re-enables propagation so the root logger renders event
output, and debug=True lifts only the DEBUG-emitting loggers
(bootstrap, migration) to DEBUG instead of setting a level on the
"kanta" parent.
State dict injection now works by parameter name (prev/state, annotation
not checked) with DictPrev/DictState tags taking precedence; DictPre/DictPost
remain as aliases. MigrationResult is renamed to MigrationReport with fields
original/version/applied; the old type alias and a deprecated .migrations
property remain. Old symbols stay covered by the original tests; new-style
tests import from the kanta root. Includes some unrelated ruff formatting.
- `@kanta.logemit` handler for completely customizable logging output, with `LogEvent` structure and `kanta.tty.Line` helper to create colorized text and fixed width fields
- `configure_logging(diff=False)` to disable diff display globally (supplementing per-transaction `logdiff=False`)
- `transaction(extra: Any = ...)` for passing extra strings or custom metadata to logs
- `@kanta.clock` to provide user controlled clock for deterministic database outputs
- Added a demo script that shows basic functions, migrations, logfmt etc.
LeoVasanko