L10n regression in 1.8.0.rc2: diffstat summary (git diff --stat, git format-patch)

L10n regression in 1.8.0.rc2: diffstat summary (git diff --stat, git format-patch)

[Peter Krefting]

Hi!

The output of "git format-patch" and "git diff --stat" no longer 
becomes localized when using 1.8.0.rc2, compared to 1.7.12

Running both versions of "git format-patch -1" from the same 
repository, with the same settings otherwise, has 1.7.12 output the 
diffstat summary in Swedish, while 1.8.0.rc2 doesn't:

   diff 1.7.12.txt 1.8.0.rc2.txt
   8c8
   <  1 fil ändrad, 1 tillägg(+)
   ---
   >  1 file changed, 1 insertion(+)
   18c18
   < 1.7.12
   ---
   > 1.8.0.rc2

The same applies to "git diff --stat" and "git show --stat". I can 
understand for "git format-patch" to default to English (but as its 
output is ignored on read, it doesn't hurt to have it localized), but 
for diff and show, it is annoying (even if the rest of the "git diff" 
output also needs to be translated eventually).

-- 
\\// Peter - http://www.softwolves.pp.se/

Re: L10n regression in 1.8.0.rc2: diffstat summary (git diff --stat, git format-patch)

[Nguyen Thai Ngoc Duy]

On Wed, Oct 17, 2012 at 5:53 PM, Peter Krefting <peter@softwolves.pp.se> wrote:
> Hi!
>
> The output of "git format-patch" and "git diff --stat" no longer becomes
> localized when using 1.8.0.rc2, compared to 1.7.12

It's the result of this discussion [1]. I don't remember exactly the
open issues. But I think it involves drawing a line between team
language vs local language, whether team language can be anything
other than English, the maintenance cost for supporting it. You're
welcome to revive the discussion. Maybe we can find a solution that is
agreed by all parties.

[1] http://thread.gmane.org/gmane.comp.version-control.git/204285/focus=204283
-- 
Duy

Re: L10n regression in 1.8.0.rc2: diffstat summary (git diff --stat, git format-patch)

[Peter Krefting]

Nguyen Thai Ngoc Duy:

> It's the result of this discussion [1]. I don't remember exactly the
> open issues. But I think it involves drawing a line between team
> language vs local language, whether team language can be anything
> other than English, the maintenance cost for supporting it

Like I said, I can see that defaulting format-patch to output English 
is a good idea (since its output is "protocol data" and not something 
for human consumption), it is the output from diff and show that I 
want to have localized.

I'll see if I can come up with a patch that cater for both use-cases. 
Maybe I can also fix so that log output is fully localized at the same 
time (translating "commit", "author", "date" and outputting locale 
dates (although my locale usess ISO dates, so I can get proper 
formatting there already)).

-- 
\\// Peter - http://www.softwolves.pp.se/

Re: L10n regression in 1.8.0.rc2: diffstat summary (git diff --stat, git format-patch)

[Peter Krefting]

> I'll see if I can come up with a patch that cater for both use-cases.

I see that I forgot to Cc you; please see the patch series starting 
with the Subject "[RFC PATCH 0/2] Localize log output", which I posted 
here yesterday.

-- 
\\// Peter - http://www.softwolves.pp.se/