Re: [PATCH][GSOC2014] changed logical chain in branch.c to lookup tables

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Mar 10, 2014, 22:39 UTC
Message-ID: <CAPig+cTOZrQY1azPCmzdrfut0S_-8ZDoob3cEqT7Z1kDpj+PQw@mail.gmail.com>
In-Reply-To: <1394488063227-7605372.post@n2.nabble.com>

On Mon, Mar 10, 2014 at 5:47 PM, Tamer TAS <tamertas@outlook.com> wrote:

Show 10 quoted lines

> Eric Sunshine wrote
>> Even allowing internationalization of them (via N_() in the location[]
>> and type[] tables) might not be sufficient since grammatical rules
>> differ from language to language.
>
> I didn't fully understand what you meant by that. Since they were being
> internationalized before using _() in the if-else bodies, wouldn't it
> produce
> the same output if I were to use the same method for location[] and type[]
> tables?

In the original code, the person translating the text has the full context:

    "Branch %s set up to track remote ref %s."

With your revision, the translator has to translate standalone words "local", "remote", "branch", "ref" without context about how they are used. He then has to translate:

    "Branch %s set up to track %s %s %s."

without context of the words being inserted, which, depending upon grammatical rules of the language can result in a different (and perhaps worse) translation than the original full-context translation.

Section 4.3 of the GNU gettext manual [1] explains the issues in more detail. I urge you to read it. The upshot is that translators fare best when handed full sentences.

Note also that your change effectively reverts d53a35032a67 [2], which did away with the sort of string composition used in your patch.

[1]: http://www.gnu.org/software/gettext/manual/gettext.html#Preparing-Strings [2]: https://github.com/git/git/commit/d53a35032a67fde5b59c8a6a66e0466837cbaf1e