Add fend to comparison table

[yonas]

Add Fend to comparison table.

[yonas]

I noticed a performance difference. Fend is quite fast:

$ hyperfine 'echo 1 ft to cm | numbat' 'echo 1 ft to cm | fend'

Benchmark 1: echo 1 ft to cm | numbat
  Time (mean ± σ):     562.6 ms ±  47.3 ms    [User: 98.9 ms, System: 9.3 ms]
  Range (min … max):   495.2 ms … 622.7 ms    10 runs
 
Benchmark 2: echo 1 ft to cm | fend
  Time (mean ± σ):       3.4 ms ±   0.1 ms    [User: 1.9 ms, System: 2.3 ms]
  Range (min … max):     3.2 ms …   4.1 ms    572 runs
 
  Warning: Command took less than 5 ms to complete. Note that the results might be inaccurate because hyperfine can not calibrate the shell startup time much more precise than this limit. You can try to use the `-N`/`--shell=none` option to disable the shell completely.
 
Summary
  echo 1 ft to cm | fend ran
  163.32 ± 14.54 times faster than echo 1 ft to cm | numbat

[sharkdp]

Do you use a release build of Numbat? 560 ms seems very slow. I see something like 135 ms (which is still quite slow, maybe we should investigate why the REPL is much slower than -e "…"). Also, Numbat is much faster if you use its -e "…" option:

▶ hyperfine 'numbat -e "1 ft to cm"'  
Benchmark 1: numbat -e "1 ft to cm"
  Time (mean ± σ):      33.9 ms ±   1.3 ms    [User: 25.0 ms, System: 8.8 ms]
  Range (min … max):    32.0 ms …  37.4 ms    84 runs

This is still slower then fend, but I'm not sure if that's a real-world concern. In Numbat's case, most of those 30 ms are spent parsing/type-checking/evaluating the standard library, which is written in Numbat itself. Your actual program (1 ft to cm) runs much faster than that.

However, if you have a non-toy example where Numbat is much slower than fend, please let us know.

[yonas]

Thanks @sharkdp, using -e is 6x faster than using echo on my system. 👍

Yes, I'm using a release build.