bug#32280: 26.1; FLYSPELL-BUFFER sometimes misbehaves for some input in a large enough buffer

[Artem Boldarev]

Hi,

Thanks, Agustin, this is an interesting find! I have altered my 
configuration as you suggested, and indeed I wasn't able to trigger the 
bug any more.

https://chaoticlab.io/pub/flyspell-bug/flyspell-bug-no-explicit-chars.png

Anyway, I am pretty confident that altering the configuration does not 
resolve the bug, but rather hides it. I think so because of the 
following reasons:

1. I do not see why my previous configuration, which uses explicitly 
specified characters, is wrong. It works fine when spell checking as you 
type and for smaller buffers and regions (when flyspell-large-region 
does not get called).

2. Without the fix, the above-discussed inconsistency exists between how 
flyspell works when you use it for:

a) spell checking as you type and checking smaller regions of text (when 
flyspell-small-region gets called).
b) spell checking large regions of text.

It should be noted, that putting the cursor on a wrongly highlighted 
word when the flyspell mode is active, gets the word rechecked and 
un-highlighted. This behaviour demonstrates this inconsistency one more 
time.

In fact, this inconsistency is the bug which I reported.

Regards,
Artem


> Hi,
>
> I am using your example file with emacs -Q (Emacs 25) and had no problems,
> flyspell-buffer works as expected after setting ispell-program-name to
> hunspell and set ru_RU as dict (hunspell 1.6.2 here).
>
> Local Ispell dictionary set to ru_RU
> Starting new Ispell process hunspell with ru_RU dictionary...
> Checking region...
> Spell Checking...100% [laborum]
> Spell Checking completed.
>
> The auto-detected values for hunspell ru_RU are
>
> (ru_RU [[:alpha:]] [^[:alpha:]]  t (-d ru_RU) nil utf-8)
>
> Wonder why otherchars is not shown.
>
> > The codepage I specified in the configuration, as it seems, is not the
> > problem as spell checking works fine *most* of the time. I could spellcheck
> > large amounts of text without any issues. It seems that hunspell always uses
> > utf-8 internally, but I am not sure: I will try to investigate this.
> Seems I was wrong, it is a long time since I digged there.
>
> ¿What happens if you use "[[:alpha:]]" and "[^[:alpha:]]" instead if the
> explicit character strings 
> "[АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЬЫЪЭЮЯабвгдеёжзийклмнопрстуфхцчшщьыъэюя]"
> and "[^АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЬЫЪЭЮЯабвгдеёжзийклмнопрстуфхцчшщьыъэюя]"?
>
> Regards,