Hyphenation of the Estonian word "näidislahendused" incorrect

[piiskop]

I was looking for how to show the hyphenation locations and found a program:

The input:

\documentclass{report}
\usepackage[estonian]{babel}
\def\rehbox{\unskip\unpenalty\setbox2\lastbox\ifhbox2
	\setbox0\hbox{\hbox{\unhbox2} \unhbox0}\expandafter\rehbox\fi}
\newcommand\printhyphens[1]{%
	\setbox0\vbox{{\setbox0\hbox{}%
			\pretolerance-1\hsize=0pt\hfuzz=\maxdimen
			\noindent\hspace*{0pt}#1\par\rehbox\unhbox0}\par}%
	\unvbox 0
}

\begin{document}
	\printhyphens{näidislahendused} 
\end{document}

The output:

näi- dis- la- hen- dus- ed

The expected output:

näi- dis- la- hen- du- sed

Please follow the rule:

2. Üksik kaashäälik täishäälikute vahel kuulub järgmisse silpi: ko-ju, du-ši, Lii-na.

[mnater]

I'm not sure, but this seems to be an issue with babel:
The estonian patterns are computed with righthyphenmin = 3

tex-hyphen/hyph-utf8/tex/generic/hyph-utf8/patterns/tex/hyph-et.tex

Line 49 in ecf976a

% right: 3

but babel uses righthyphenmin = 2
https://github.com/latex3/babel/blob/b488d60c6b12eefc664077666b4e207f90b63889/locale/et/babel-et.ini#L152 .

If righthyphenmin was set to 3 (as requested by the patterns) the word would hyphenate to näi•dis•la•hen•dused missing the last hyphenation opportunity but being correct.

[jbezos]

@mnater The ‘hyphenmins’ in the hyphenation files doesn’t necessarily reflect the values set in patgen to generate them. This is particularly true in rule-based patterns (ie, created without patgen), like those for spanish, which sets the hyphenmins to 2/2, the ‘typographical’ limit, as opposed to the ‘technical’ limit (which is 1/1). Please, also note the original babel style and the patterns are the work of one person, so it’s doubtful there is a mistake here (btw, polyglossia also sets 2/2).