(please ignore my last 3 messages - I was just me having trouble sending this message - here is the complete message)

>now I'd suggest this one as more readable:
>
>    nplurals=3; plural=n==1 ? 0 : n==2 ? 2 : 1;
>
>Niv, is this order of the translated strings what you want:
>
>msgstr[0] "Singular form"
>msgstr[1] "Plural form (including zero)"
>msgstr[2] "Dual form"
>
>Can we have this formula for now?

That's great, thank you.

>Now it's not clear to me what you mean about the "many" form.
>You wrote that it is correct not to use this form but it would
>be nicer to have it. If it would be nice then let's be nice,
>using more plural forms is not beyond our abilities!

as using the regular plural form is correct (seem natural to the user*),
it will only add unneeded extra complexity,
so we'd prefer to leave it for now.
meanwhile,
projects who want can do it manipulating the .po file header (as GNOME Software did)
for overall standard,
we can always turn over the world again at later time :)

>But for now I don't understand where the "many" form should be used:
>* for the numbers: 20, 30, 40, ..., 100, 110, 120, ...
>or
>* for the numbers: 11, 12, 13, ..., 99, 100, 101, ...
the latter, for the numbers: 11, 12, 13, ..., 99, 100, 101, ... (n>10)
but again, using the regular plural form is correct*.

Thanks,
Niv

*user reading in hebrew the sentence "2 days ago" translated as "לפני 2 ימים" will get the filling this sentence is not natural and should be "לפני יומיים"

 user reading in hebrew the sentence "20 days ago" translated as "לפני 20 ימים" will continue reading without paying attention that the form "לפני 20 יום" exists as well



2016-09-15 2:19 GMT+03:00 Rafal Luzynski <digitalfreak@lingonborough.com>:
14.09.2016 13:30 ניב באר <bloop93@gmail.com> wrote:
>
>
>  Just to make sure,
>  We are requesting to add the rule for dual form.. the other rule can be
> ignored.
>
>  0 : n==1 : singular form
>  1 : n==0||n >2 :plural form
>  2 : n==2 : dual form
>
>  We do not actually need the many form for the translation to be correct.
>
>  (Assuming putting dual form on 2 can mainyain some bacward compatibility)
>
>  Thanks,
>  Niv

As for now it looks clear to me that Hebrew language needs at least
the dual form to be added. Even if dual form usually will be the
same as the plural we need a mechanism and leave it to the translators
whether they use it in particular cases or not.

Now it's not clear to me what you mean about the "many" form.
You wrote that it is correct not to use this form but it would
be nicer to have it. If it would be nice then let's be nice,
using more plural forms is not beyond our abilities! But for
now I don't understand where the "many" form should be used:

* for the numbers: 20, 30, 40, ..., 100, 110, 120, ...
or
* for the numbers: 11, 12, 13, ..., 99, 100, 101, ...

As long as it's clear that dual number is necessary and it's
not clear whether and when the "many" number is necessary
I suggest adding the dual number rule and not adding the "many"
number yet. As far as I understood, missing dual number is
incorrect while missing "many" number is not.

Although previously I referred to this formula: [1]

    nplurals=3; plural=n==1 ? 0 : (n>2||n==0) ? 1 : 2;

now I'd suggest this one as more readable:

    nplurals=3; plural=n==1 ? 0 : n==2 ? 2 : 1;

Niv, is this order of the translated strings what you want:

msgstr[0] "Singular form"
msgstr[1] "Plural form (including zero)"
msgstr[2] "Dual form"

Can we have this formula for now?

For the record, here [2] is a formula for Hebrew according to
what CLDR curently says:

    nplurals=4; plural=(n == 1) ? 0 : ((n == 2) ? 1 : (((n < 0 || n > 10) && n %
10 == 0) ? 2 : 3))

Regards,

Rafal


[1] https://git.gnome.org/browse/gnome-software/tree/po/he.po
[2] http://mlocati.github.io/cldr-to-gettext-plural-rules/