Re: [ANN] luaRT 0.9 beta1

[samir.tine@...]

Le 2021-01-19 19:31, samir.tine@luart.org a écrit :
> 1) Yes %w pattern will always match what you call "national alphabet
> characters". That maintains compatibility with standard Lua.
>
> 2) I have tested string.find("Ё", "[А-Я]") in LuaRT : it returns
> correctly 1,2 that means the Ё has been correctly found ?
>
> Regards,
>
> Samir
>
> 19 janvier 2021 07:28 "Egor Skriptunoff" <egor.skriptunoff@gmail.com>
> a écrit:
>
> > On Mon, Jan 18, 2021 at 10:17 PM wrote:
> >
> > > > I see you have removed os.setlocale() from the standard library.
> > > > How to switch between C locale and national locale?
> > > LuaRT does not use the concept of C locale.
> >
> > Does it mean that the "%w" pattern will always match national
> > alphabet characters?
> > In some projects I prefer to have "%w" as ASCII7-only.
> > Do I have to use "[A-Za-z%d]" instead in LuaRT?
> >
> > > Char intervals are treated the same. An accentued character (for
> > > example
> > > "à") is not in the interval [A-Z] (which includes letters, but
> > > not
> > > accentued letters).
> >
> > Is char interval interpreted from the point of view of the alphabet
> > or from the point of view of codepoint numeric values?
> > For example, "А" is the first letter of the Russian alphabet, "Я"
> > is the last letter.
> > The letter "Ё" is somewhere between them in the alphabet,
> > but codepoint("Ё") is beyond the numeric range of
> > codepoint("А")...codepoint("Я")
> > So, what would be the result of string.find("Ё", "[А-Я]") in
> > LuaRT?

In fact, it returns 1, 1 (matching the right character).
A bug consisting in a miscalculation of the end of the match was found 
with your help :)

Thank you !