Re: [PATCH v2 18/22] tests: use test_i18n* functions to suppress false positives

From: Vasco Almeida <vascomalmeida@sapo.pt>
Date: May 27, 2016, 19:34 UTC
Message-ID: <5748A13A.6070608@sapo.pt>
In-Reply-To: <xmqqd1o7sa30.fsf@gitster.mtv.corp.google.com>

Às 17:08 de 27-05-2016, Junio C Hamano escreveu:

Show 20 quoted lines

> Vasco Almeida <vascomalmeida@sapo.pt> writes:
> 
>> diff --git a/t/t1307-config-blob.sh b/t/t1307-config-blob.sh
>> index 3c6791e..4079fef 100755
>> --- a/t/t1307-config-blob.sh
>> +++ b/t/t1307-config-blob.sh
>> @@ -64,7 +64,7 @@ test_expect_success 'parse errors in blobs are properly attributed' '
>>  
>>  	# just grep for our token as the exact error message is likely to
>>  	# change or be internationalized
>> -	grep "HEAD:config" err
>> +	test_i18ngrep "HEAD:config" err
>>  '
> 
> It is unfortunate that the gettext-poison mechanism is too dumb to
> notice that it is clobbering a format string with placeholders and
> leave them intact, which is what the comment above this change is
> wishing for.  I do not think we will be granting that wish any time
> soon, so perhaps remove the two lines while we are at it?
>

Yes, that was more or less what I thought when read that comment, but forgot about it after. I'll remove that comment in the next re-roll. Also, because it can be deceiving for someone not entirely sure how does gettext poison, translations and tests work together when test are run.

For the record, test are run under C locale, so translations don't matter in tests. Under GETTEXT_POISON build, all strings interpolated by gettext (for instance _("a message from git") in C language) are replaced by garbage string "# GETTEXT POISON #".