Re: [PATCH v3 00/15] Mark "init", "clone", and "add" output for translation

From: Junio C Hamano <gitster@pobox.com>
Date: Feb 22, 2011, 18:11 UTC
Message-ID: <7vhbbwdjnm.fsf@alter.siamese.dyndns.org>
In-Reply-To: <20110221101229.GA32137@elie>

Jonathan Nieder <jrnieder@gmail.com> writes:

Show 14 quoted lines

> This series still includes the skeleton gettext.h wrapper.
>
> Changes from v1:
>
>  - clearer commit messages, hopefully
>  - clearer documentation for GETTEXT_POISON in the makefile
>
> Please feel free to try it out, complain, ack, send patches, and so
> on.
>
> Jonathan Nieder (2):
>   i18n: do not poison translations unless GIT_GETTEXT_POISON envvar is
>     set
>   update-index --refresh --porcelain: add missing const

I took this series (minus 4/15), which roughly corresponds to up to 13/72 of Ævar's original. A few quick comments:

 * 01/15
   Thanks for guarding the header file from double inclusion.  I have this
   nagging feeling that the inline _() function might be too clever for
   some compilers (non gcc and perhaps older versions of gcc), but we can
   always ifdef it away if it turns problematic.

 * 02/15
   I am not entirely happy that it is a too generic name to contaminate
   namespace of programs that may want to get i18n support while wanting
   to catch bugs with their own unrelated poisoning scheme.

   As to Malkovich, I tend to agree with your reasoning that code that
   emit messages prefixed by "#" rebase relies on should be restructured
   to limit the scope of translation to just the body of these lines, and
   that should be done in a separate patch.

 * 04/15
   Deferred per request from Ævar.