vim-patch:9.2.1103: regexp compilation decodes ASCII bytes as multi-byte

Problem:  When compiling a regexp the lexer calls mb_ptr2char() and
          utf_ptr2len() for every character, including plain ASCII, even
          though a byte below 0x80 is always a single character equal to
          itself. Patterns are mostly ASCII, so this is wasted work on the
          hot compile path.
Solution: In peekchr() and skipchr() take a fast path for bytes below 0x80,
          skipping the multi-byte decode calls (Julien Voisin).

peekchr() already holds regparse[0] in "curchr", so a byte below 0x80 needs no
mb_ptr2char() call; the backslashed case is handled the same way. skipchr()
likewise advances by one byte directly instead of calling utf_ptr2len(). Both
regexp engines share this lexer, so both compile faster.

NFA pattern compilation is about 5% cheaper (skipchr() -29%, peekchr() -13%);
loading a set of syntax files is about 1.4% faster end to end.

closes: vim/vim#21301

https://github.com/vim/vim/commit/e5da1f1f037d2549c883db709ecde2928d0f047e

Co-authored-by: Julien Voisin <julien.voisin@dustri.org>

0 files changed, 0 insertions, 0 deletions