# [PATCH] t5510: become resilient to GETTEXT_POISON 6 messages from 2016-07-25 to 2016-07-26. Participants: Vasco Almeida, Junio C Hamano, Duy Nguyen, Ævar Arnfjörð Bjarmason. Thread: https://gitlist.dev/t/42917 ## Vasco Almeida, 2016-07-25 09:31 Subject: [PATCH] t5510: become resilient to GETTEXT_POISON Message-ID: <1469439068-2724-1-git-send-email-vascomalmeida@sapo.pt> URL: https://gitlist.dev/e/1469439068-2724-1-git-send-email-vascomalmeida%40sapo.pt ``` Replace gettext poison text with appropriate values to be able to cut the right output of git fetch command for comparison. The first gettext poison falls from the previous line into the next because the poison does not add a newline, so we must replace it with nothing. Signed-off-by: Vasco Almeida --- t/t5510-fetch.sh | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh index 6bd4853..b261223 100755 --- a/t/t5510-fetch.sh +++ b/t/t5510-fetch.sh @@ -694,7 +694,10 @@ test_expect_success 'fetch aligned output' ' ( cd full-output && git -c fetch.output=full fetch origin 2>&1 | \ - grep -e "->" | cut -c 22- >../actual + grep -e "->" | \ + sed -e "/master/ s/# GETTEXT POISON #//" \ + -e "/tag/ s/# GETTEXT POISON #/[new tag] /" | \ + cut -c 22- >../actual ) && cat >expect <<-\EOF && master -> origin/master @@ -709,7 +712,10 @@ test_expect_success 'fetch compact output' ' ( cd compact && git -c fetch.output=compact fetch origin 2>&1 | \ - grep -e "->" | cut -c 22- >../actual + grep -e "->" | \ + sed -e "/master/ s/# GETTEXT POISON #//" \ + -e "/extraa/ s/# GETTEXT POISON #/[new tag] /" | \ + cut -c 22- >../actual ) && cat >expect <<-\EOF && master -> origin/* -- 2.7.4 ``` ## Junio C Hamano, 2016-07-25 15:16 Subject: Re: [PATCH] t5510: become resilient to GETTEXT_POISON Message-ID: URL: https://gitlist.dev/e/CAPc5daVV8EkMXWeTiZN9bEXXAejmbDbT9L7SNCc4jNtXYVomLQ%40mail.gmail.com In-Reply-To: <1469439068-2724-1-git-send-email-vascomalmeida@sapo.pt> ``` On Mon, Jul 25, 2016 at 2:31 AM, Vasco Almeida wrote: > Replace gettext poison text with appropriate values to be able to cut > the right output of git fetch command for comparison. Hmm, as these tests are _all_ about human-readable output, it probably is sufficient to skip them using prerequiste, I would think. We do not want each individual test to have too intimate knowledge on how POISON strings look like. ``` ## Duy Nguyen, 2016-07-25 16:07 Subject: Re: [PATCH] t5510: become resilient to GETTEXT_POISON Message-ID: URL: https://gitlist.dev/e/CACsJy8BLTdsUqN-jUXUmX%2B9npja9H5rO8svM18YptrxtJ7W64w%40mail.gmail.com In-Reply-To: ``` On Mon, Jul 25, 2016 at 5:16 PM, Junio C Hamano wrote: > On Mon, Jul 25, 2016 at 2:31 AM, Vasco Almeida wrote: >> Replace gettext poison text with appropriate values to be able to cut >> the right output of git fetch command for comparison. > > Hmm, as these tests are _all_ about human-readable output, it probably is > sufficient to skip them using prerequiste, I would think. We do not want each > individual test to have too intimate knowledge on how POISON strings look > like. Yeah these tests are about alignment, they are probably useless anyway after the text is poisoned (and has the same length). -- Duy ``` ## Vasco Almeida, 2016-07-26 12:58 Subject: [PATCH] t5510: skip tests under GETTEXT_POISON build Message-ID: <1469537934-16676-1-git-send-email-vascomalmeida@sapo.pt> URL: https://gitlist.dev/e/1469537934-16676-1-git-send-email-vascomalmeida%40sapo.pt In-Reply-To: <1469439068-2724-1-git-send-email-vascomalmeida@sapo.pt> ``` Skip tests when running under GETTEXT_POISON build and run them with C_LOCALE_OUTPUT prerequisite. These tests are irrelevant under GETTEXT_POISON because they test text output alignment which GETTEXT_POISON turns useless. Signed-off-by: Vasco Almeida --- t/t5510-fetch.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh index 6bd4853..668c54b 100755 --- a/t/t5510-fetch.sh +++ b/t/t5510-fetch.sh @@ -688,7 +688,7 @@ test_expect_success 'fetching with auto-gc does not lock up' ' ) ' -test_expect_success 'fetch aligned output' ' +test_expect_success C_LOCALE_OUTPUT 'fetch aligned output' ' git clone . full-output && test_commit looooooooooooong-tag && ( @@ -703,7 +703,7 @@ test_expect_success 'fetch aligned output' ' test_cmp expect actual ' -test_expect_success 'fetch compact output' ' +test_expect_success C_LOCALE_OUTPUT 'fetch compact output' ' git clone . compact && test_commit extraaa && ( -- 2.7.4 ``` ## Junio C Hamano, 2016-07-26 16:53 Subject: Re: [PATCH] t5510: skip tests under GETTEXT_POISON build Message-ID: URL: https://gitlist.dev/e/xmqq4m7c2ve8.fsf%40gitster.mtv.corp.google.com In-Reply-To: <1469537934-16676-1-git-send-email-vascomalmeida@sapo.pt> ``` Vasco Almeida writes: > Skip tests when running under GETTEXT_POISON build and run them with > C_LOCALE_OUTPUT prerequisite. > > These tests are irrelevant under GETTEXT_POISON because they test text > output alignment which GETTEXT_POISON turns useless. > > Signed-off-by: Vasco Almeida > --- > t/t5510-fetch.sh | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh > index 6bd4853..668c54b 100755 > --- a/t/t5510-fetch.sh > +++ b/t/t5510-fetch.sh > @@ -688,7 +688,7 @@ test_expect_success 'fetching with auto-gc does not lock up' ' > ) > ' > > -test_expect_success 'fetch aligned output' ' > +test_expect_success C_LOCALE_OUTPUT 'fetch aligned output' ' > git clone . full-output && > test_commit looooooooooooong-tag && > ( > @@ -703,7 +703,7 @@ test_expect_success 'fetch aligned output' ' > test_cmp expect actual > ' > > -test_expect_success 'fetch compact output' ' > +test_expect_success C_LOCALE_OUTPUT 'fetch compact output' ' > git clone . compact && > test_commit extraaa && > ( Makes sense, will queue. This is a tangent, but it may make sense for us to start thinking about retiring one of the two prerequisites, GETTEXT_POISON and C_LOCALE_OUTPUT. Back when 5e9637c6 (i18n: add infrastructure for translating Git with gettext, 2011-11-18) introduced the former, test_have_prereq did not support a negated prerequisite, so the commit added GETTEXT_POISON prerequisite; if we had the modern test_have_prereq, we would have written test_expect_success GETTEXT_POISON '...' that appear in t0205 as test_expect_success !C_LOCALE_OUTPUT '...' I would think. ``` ## Ævar Arnfjörð Bjarmason, 2016-07-26 20:11 Subject: Re: [PATCH] t5510: skip tests under GETTEXT_POISON build Message-ID: URL: https://gitlist.dev/e/CACBZZX4r5jax62oVYFsWONkyREshipGC8AFu8yfr6B7y9bMQDQ%40mail.gmail.com In-Reply-To: ``` On Tue, Jul 26, 2016 at 6:53 PM, Junio C Hamano wrote: > [...] Back when 5e9637c6 (i18n: add infrastructure for > translating Git with gettext, 2011-11-18) introduced the former, > test_have_prereq did not support a negated prerequisite, so the > commit added GETTEXT_POISON prerequisite; if we had the modern > test_have_prereq, we would have written > > test_expect_success GETTEXT_POISON '...' > > that appear in t0205 as > > test_expect_success !C_LOCALE_OUTPUT '...' > > I would think. Maybe the names of the test prerequisites should be merged. I can't think of a rea As for the GETTEXT_POISON facility in general, I haven't worked much if at all on the i18n toolchain since I initially wrote the gettext support so I think at this point it's for others to say whether stuff like this is useful. But for what it's worth the v1.7.4.1-65-gbb946bb commit explains better what it's for: This is a debugging aid for people who are working on the i18n part of the system, to make sure that they are not marking plumbing messages that should never be translated with _(). I.e. so the person gettext-izing something can actively spot issues with marking strings for translations right away. ```