Re: [PATCH] 5x speedup of flyspell-buffer

[Ilya Zakharevich <nospam-abuse@ilyaz.org>]

On Thu, Apr 13, 2023 at 09:54:59AM +0300, Eli Zaretskii wrote:
> > Date: Wed, 12 Apr 2023 06:50:30 -0700
> > From: Ilya Zakharevich <ilya@ilyaz.org>
> > 
> > With the included patch, on my system flyspell-buffer speeds up 5
> > times.  For my typical file sizes, this is 20sec -> 4sec on a 600K
> > LaTeX file with 10,000 words detected by `aspell´ — which makes
> > using flyspell feasible.
> > 
> >   The version is not the most recent, but IIUC, flyspell did not change.
> > 
> > (The number 1.05 below is chosen to lower overhead of messages 50 times,
> > from 5x to 0.1x.  It is related to log(10000)/0.05 ∼ 10000/50.)
> 
> Thanks, but does it indeed make sense to print these progress message
> with logarithmically-increasing intervals?  Why not simply decimate
> the messages by printing them every N words or so?  N could be
> computed at the beginning of the command as function of the size of
> the region to be spell-checked, assuming some suitable value of
> average length of a word.  WDYT?

Without profiling, how to find out “how much overhead is tolerable to
the user”?!

Going logarithmic avoids all these complications.  The REAL purpose
of these messages is “the feedback to the user why the UI is locked now”.

Thanks,
Ilya

P.S.  BTW, there is very little need to lock the UI.  On my (very old
      and slow) system, these 10,000 misspellings are reported in 300ms
      (even though aspell is cygwin — read “slow” — one).  After this
      list of misspellings is swallowed, the rest of marking can be
      done asynchronously, via timers.

      I may try to find more time to investigate (and maybe implement) this…