tstack/lnav v0.14.1-rc1

on GitHub

lnav v0.14.1

Features:

• Added the :reload-view command, bound to F5/⌘-R,
  that re-runs the operation that populated the current
  view.
  In the DB view, this re-executes the last SQL query; in
  the TIMELINE view, it rebuilds the index. Views that
  don't have a meaningful reload report an error.
• The DB view now shows a status bar above the bottom
  status bar with the SQL query that populated the view,
  the relative time when it was run, and how long it took.
  A reload icon on the left side of the bar can be clicked
  to re-run the query. For queries that read from log-backed
  tables, the status bar also calls out whether the results
  are "on current log data" or "on old log data" relative to
  what lnav is currently indexing. The same information is
  exposed via the new lnav_views.view_details column.
• For terminals that support the Kitty Keyboard
  protocol, the following hotkeys are now supported
  in the prompt:
  • ⌘-C to copy the current selection to the
    clipboard.
  • ⌘-A to select all text.
  • ⌘-X to cut the current selection to the
    clipboard.
  • ⌘-Z to undo the last change.
• The multi-line prompt can now be resized with the
  keyboard: ALT-= grows the prompt by one line and
  ALT-- shrinks it by one line. This complements
  the existing click-and-drag resize on the prompt's
  status bar.
• Additional readline-style key bindings in the
  prompt (issue #1676):
  • ALT-f / ALT-b move forward/backward by word.
  • ALT-d cuts to the end of the next word.
  • ALT-BACKSPACE is an alias for CTRL-W.
  • ALT-l / ALT-u lower- or upper-case the
    next word.
  • ALT-c capitalizes the next word.
  • CTRL-h and CTRL-d are aliases for
    BACKSPACE and DELETE.
  • CTRL-t transposes the two characters before
    the cursor.
• In the log message details (opened by pressing p),
  a File: row has been added that lists the file and
  line number the log message is from. Also, for JSON
  logs, you can focus on the JSON fields: row and
  press c to copy the raw log message to the
  clipboard.
• The measure_with_units collator now recognizes
  (KiB, MiB, ...).
• Added disfavors to external-editor configuration
  to express file names that the editor is not interested
  in handling.
• The lnav_views.view_details column is now populated
  for every time-based view (log, histogram, spectro,
  timeline, db) with a JSON zoom-level field reflecting
  the current :zoom-to setting.
• The :zoom-to command (and the z/Z hotkeys) now
  affects only the currently focused view; previously,
  zooming in either the histogram or the spectrogram
  view would change both views' zoom level in lockstep.
  Each view (log, histogram, spectro, timeline, db) now
  has its own independent zoom level.
• The :zoom-to command now accepts + and - as
  shortcuts to step in or out by one level from the
  current view's zoom (the same behavior as the z/Z
  hotkeys).
• SQL statements can now use $zoom_level to refer to
  the current zoom level in the DB view. Pressing
  z/Z in the DB view will now rerun the last query
  with the new zoom value. The stats.hist PRQL
  function now uses $zoom_level as the default value
  for the slice parameter.
• A stats.timeseries PRQL function has been added to
  make it easier to perform an aggregation over buckets
  of time.

Breaking changes:

• The humanize_file_size() SQLite function now
  uses 1,000 for the base instead of 1,024.

Bug Fixes:

• A PRQL query can now start with let in interactive
  mode.
• Fix a bug in file loading that could cause a short
  read and crash in some situations.
• Fix a lockup when viewing a file that contained log
  messages and lots of binary data.
• Update Regex101 import functionality to handle a
  change in their API response.
• More hardening for some diabolical inputs:
  • Unsupported escape-sequences were ignored before,
    but they are displayed now.
  • Checks for archives with file paths that could
    escape containment.
• The duplicate file check is less aggressive now.
  Previously, if the first lines of logfiles matched
  exactly, they were considered duplicates and the
  smallest/oldest was hidden. Now, the duplication
  check is only done on files that contain at least
  100 lines and those lines are checked to see if
  they have the same timestamp/file-offset.
• In the TIMELINE view, tags whose names start or end
  with "stop", "stopped", "end", "ended", "finish", or
  "finished" (case-insensitive) are now paired with
  matching "start", "started", or "begin" tags that
  share the same base name (for example, #start-foo
  is paired with #stop-foo). The start tag's row
  spans from the start time to the stop time, and the
  stop tag is no longer shown as a separate row. Tags
  without a base name (such as a bare #start) are
  not paired.