Re: WIN32 pg_import_system_collations

On Tue, Jan 25, 2022 at 11:40 AM Peter Eisentraut <
peter.eisentraut@enterprisedb.com> wrote:

> On 24.01.22 22:23, Dmitry Koval wrote:
>

Thanks for looking into this.


> > +/*
> > + * Windows will use hyphens between language and territory, where POSIX
> > + * uses an underscore. Simply make it POSIX looking.
> > + */
> > + hyphen = strchr(localebuf, '-');
> > + if (hyphen)
> > +    *hyphen = '_';
> >
> > After this block modified collation name is used in function
> >
> > GetNLSVersionEx(COMPARE_STRING, wide_collcollate, &version)
> >
> > (see win32_read_locale() -> CollationFromLocale() -> CollationCreate()
> > call). Is it correct to use (wide_collcollate = "en_NZ") instead of
> > (wide_collcollate = "en-NZ") in GetNLSVersionEx() function?
>

The problem that David Rowley addressed was coming from Windows collations
in the shape of "English_New Zealand", GetNLSVersionEx() will work with
both "en_NZ" and "en-NZ". You can check collversion in pg_collation in the
patched version.

>
> I don't really know if this is necessary anyway.  Just create the
> collations with the names that the operating system presents.  There is
> no requirement to make the names match POSIX.
>
> If you want to make them match POSIX for some reason, you can also just
> change the object name but leave the collcollate/collctype fields the
> way they came from the OS.
>

I think there is some value in making collation names consistent across
different platforms, e.g. making user scripts more portable. So, I'm doing
that in the attached version, just changing the object name.

Regards,

Juan José Santamaría Flecha