v8.0.0

[8.0.0] - 05.08.2026

Breaking Changes

• The following types and functions have been removed:

  • TableComponent enum
  • The modifiers module and Table::apply_modifier have been removed.
  • Table::current_style_as_preset
  • Table::set_style
  • Table::style
  • Table::remove_style
  • Table::load_preset

• The positional preset string format has been replaced with the new TableStyle, LineStyle and ContentLineStyle types.
  The old approach of having "stringified" presets was too finicky and also implicitly bound to the order of variants in the old TableComponent enum.
  It was just not a good design.

  The new design, in comparison, has a nice and (hopefully) intuitive API to define a style.
  On top of that, all functions on TableStyle are const, so custom styles can still be declared as constants:

  const MY_STYLE: TableStyle = TableStyle::new()
      .top_border(LineStyle::new('┌', '─', '┬', '┐'))
      .header_lines(ContentLineStyle::new('│', '┆', '│'))
      .header_separator(LineStyle::new('╞', '═', '╪', '╡'))
      .content_lines(ContentLineStyle::new('│', '┆', '│'))
      .bottom_border(LineStyle::new('└', '─', '┴', '┘'));

  The new API looks as follows:

  • Presets in comfy_table::presets::* are now TableStyle constants.
  • Table::load_style(style) has been added to load a preset or custom style.
  • Table::style() and Table::style_mut() to get a handle to the style used by the table.
    The mut handle can be used to change the style.
  • Instead of modifiers, there are now functions on TableStyle TableStyle::with_rounded_corners/TableStyle::with_solid_inner_borders.
  • All fields of TableStyle, LineStyle and ContentLineStyle are public:

    table.style_mut().top_border.left = Some('╭');

• The default truncation indicator is now … instead of ....
  This should make it more obvious that there's more text and consumes less visual space.

Fix

• Various performance improvements that reduce table formatting time quite a bit:

  • ~36% for "normal" tables (The readme of the project). 17µs -> 11µs
  • ~34% for "larger" tables, which I would consider a reasonable usecase. 253µs -> 172µs
  • ~70% for very large tables, i.e. a 10x500 table. -> 20.8ms -> 6.15ms

  Most of these improvements boil down to:

  • Less string allocations.
  • Special handling when handling ASCII-only text, as grapheme handling can be skipped in those cases.
  • Moving computational heavy stuff (such as UTF-8 width calculations) out of hot loops.

• Fixed a non-critical bug that resulted in less-than optimal layouting when UTF-8 characters were involved.

• No longer panic when rendering a Table that contains a row with Row::max_height(0).

• Don't print multi-width truncation indicators, if they would be longer than the column.