I found that set ignore-case = smart-case does not work when searching a term that includes the space character, e.g., searching 'foo bar' does not match 'Foo Bar'.

• This is because the category for space is UTF8PROC_CATEGORY_ZS, which has the value 23, aka 0x17, or 0b10111, and the smart-case test, utf8_string_contains_uppercase, which calls utf8_string_contains, is using bitwise AND to check the category:

   if (property->category & category)
  

  So then the space character -- category 0x17 -- matches the uppercase category, UTF8PROC_CATEGORY_LU = 1. I.e., 0x17 & 1 is true.

• My solution is to change the bitwise AND to the equality operator, ==.

  This is based on my understanding that the utf8proc categories are not bitmask values (and that utf8proc_get_property returns a single value, and not a bitmask), although I could be mistaken (this is my first time examining the utf8proc library).

• This regression occurred in 29524d0.

Oh, and by the way, tig is such a delight to use! I discovered it when I got a MacBook for work and found out how slow gitk runs outside of Linux, and I haven't used gitk since -- tig's gotta be one of my favorite developer tools. Great work!!