Re: Emacs i18n

[Paul Eggert]

Richard Stallman wrote:
> Compare this
>
>    (numeric-case NUMBER
>        (russian-masc "%d байт скопирован, %s, %s")
>        (russian-fem "%d байта скопировано, %s, %s")
>        (russian-neut "%d байт скопировано, %s, %s"))
>
> with this:
>
>      "Plural-Forms: nplurals=3; plural=(n%10==1 && n%100!=11 ? 0 : n%10>=2 && 
> n%10<=4
>      && (n%100<10 || n%100>=20) ? 1 : 2);\n"
>      ...
>      #: src/dd.c:822
>      #, c-format
>      msgid "%<PRIuMAX> byte copied, %s, %s"
>      msgid_plural "%<PRIuMAX> bytes copied, %s, %s"
>      msgstr[0] "%<PRIuMAX> байт скопирован, %s, %s"
>      msgstr[1] "%<PRIuMAX> байта скопировано, %s, %s"
>      msgstr[2] "%<PRIuMAX> байт скопировано, %s, %s"

I'm afraid that's not a apples-to-apples comparison. The first form contains 
only the Russian translations, whereas the second form contains much more 
information: the source-code location of the untranslated strings, a copy of the 
untranslated English-language strings, and the general rules for Russian (the 
last is shared among all the Russian translations, not just the translations 
listed here). This extra information is useful for translators, and it has a 
reasonably extensive software suite that already supports it, not to mention 
translators who are already used to it.

> I can envision something like this:
>
>        "russian-nom:%d байт%| скопирован%|, %s, %s"
>
> where the 'russian-nom' operator would replace the two %| sequences
> with the appropriate declensional suffixes for the nominative case.

But Russian declension is not that simple. The Russian word for "byte" is 
"байт", but its plural form depends not only on the number (as in the above 
examples) but also in its case: the "байт" and "байта" in the above examples are 
not exhaustive. And some words have irregular declensions: for example, ребёнок 
(singular) versus де́ти (plural) for the same noun. And it's not just nouns and 
pronouns that are affected: adjectives also have singular and plural forms. And 
I have by no means exhausted the issues involved here; to get a better feeling 
for the complexity in this area, please see:

https://en.wikipedia.org/wiki/Russian_declension

Although it wouldn't be impossible for Emacs Lisp code to handle all the special 
cases for Russian declension, it would be tricky to implement, or to document it 
in a way that translators would easily understand. And we'd also have to 
implement and document similarly tricky rules for other languages. And we'd have 
to deal with the fact that not every Russian-speaker agrees with how to decline 
words like "байт" that are imported from English. These sorts of issues should 
be delegated to translators, not to likely-fragile code in Emacs Lisp (a 
technology that translators typically do not grok).

In contrast, the gettext way is relatively simple and easily understood, and is 
already common practice.