On Friday, February 16, 2024 at 4:33:16 AM UTC-6 Edward K. Ream wrote: > Experiments show that it would be difficult/impossible to make Leo's new beautifier 10x faster!
Here are the timing statistics to beautify leoAst.py with Leo's new beautifier: tbo: 0.1 sec. dirty: 0 checked: 1 beautified: 0 in leo\core\leoAst.py 0.1 seconds is 100 milliseconds, so to be 10x faster, the total *budget* for the faster beautifier would be only 10 milliseconds! However, a *Nim-only* prototype takes 3 milliseconds to *approximate* tokenizing leoAst.py. A full Nim-only tokenizer surely would take significantly longer. Finally, the following nimpy code takes about 500(!!) milliseconds to tokenize leoAst.py. let text_stream = io.StringIO(contents) let readline = text_stream.readline let tokens = tokenize_module.generate_tokens(readline) Sure, I may have made a newbie nimpy mistake, but it appears highly unlikely that nimpy can make Python's tokenizer module available at a reasonable speed. *Summary* Leo's new beautifier is surprisingly fast, probably because Python's tokenizer runs at C speed. As a result, it would be difficult/impossible (in any language) to get a 10x speedup. Edward -- You received this message because you are subscribed to the Google Groups "leo-editor" group. To unsubscribe from this group and stop receiving emails from it, send an email to leo-editor+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/leo-editor/6ffbe12c-fa91-467b-b261-06efb078c957n%40googlegroups.com.
