• Zbigniew Jędrzejewski-Szmek's avatar
    diff --stat: use less columns for change counts · dc801e71
    Zbigniew Jędrzejewski-Szmek authored
    Number of columns required for change counts is now computed based on
    the maximum number of changed lines instead of being fixed. This means
    that usually a few more columns will be available for the filenames
    and the graph.
    
    The graph width logic is also modified to include enough space for
    "Bin XXX -> YYY bytes".
    
    If changes to binary files are mixed with changes to text files,
    change counts are padded to take at least three columns. And the other
    way around, if change counts require more than three columns, then
    "Bin"s are padded to align with the change count. This way, the +-
    part starts in the same column as "XXX -> YYY" part for binary files.
    This makes the graph easier to parse visually thanks to the empty
    column. This mimics the layout of diff --stat before this change.
    
    Tests and the tutorial are updated to reflect the new --stat output.
    This means either the removal of extra padding and/or the addition of
    up to three extra characters to truncated filenames. One test is added
    to check the graph alignment when a binary file change and text file
    change of more than 999 lines are committed together.
    Signed-off-by: Zbigniew Jędrzejewski-Szmek's avatarZbigniew Jędrzejewski-Szmek <zbyszek@in.waw.pl>
    Signed-off-by: default avatarJunio C Hamano <gitster@pobox.com>
    dc801e71
t7602-merge-octopus-many.sh 1.95 KB