# [PATCH/RFC] t0060: basename("//") is implementation defined 4 messages from 2016-01-13 to 2016-01-14. Participants: tboegi@web.de, Junio C Hamano, Johannes Schindelin, Torsten Bögershausen. Thread: https://gitlist.dev/t/41185 ## tboegi@web.de, 2016-01-13 20:56 Subject: [PATCH/RFC] t0060: basename("//") is implementation defined Message-ID: <1452718561-32226-1-git-send-email-tboegi@web.de> URL: https://gitlist.dev/e/1452718561-32226-1-git-send-email-tboegi%40web.de ``` From: Torsten Bögershausen From http://pubs.opengroup.org/onlinepubs/9699919799/functions/basename.html basename("//") may return either "//" or "/". Linux returns "/", some systems like Mac OS X return "//". Skip the test unless NO_LIBGEN_H is set and gitbasename() from compat/basename is used. Signed-off-by: Torsten Bögershausen --- My version of a hot fix test-path-utils.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/test-path-utils.c b/test-path-utils.c index 4ab68ac..db0bb39 100644 --- a/test-path-utils.c +++ b/test-path-utils.c @@ -142,7 +142,9 @@ static struct test_data dirname_data[] = { { ".", "." }, { "..", "." }, { "/", "/" }, +#if defined(NO_LIBGEN_H) { "//", "//" }, +#endif #if defined(__CYGWIN__) && !defined(NO_LIBGEN_H) { "///", "//" }, { "////", "//" }, -- 2.7.0.278.g9379adb.dirty ``` ## Junio C Hamano, 2016-01-13 21:16 Subject: Re: [PATCH/RFC] t0060: basename("//") is implementation defined Message-ID: URL: https://gitlist.dev/e/xmqqegdlmb7f.fsf%40gitster.mtv.corp.google.com In-Reply-To: <1452718561-32226-1-git-send-email-tboegi@web.de> ``` tboegi@web.de writes: > From: Torsten Bögershausen > > From > http://pubs.opengroup.org/onlinepubs/9699919799/functions/basename.html > basename("//") may return either "//" or "/". > > Linux returns "/", some systems like Mac OS X return "//". > Skip the test unless NO_LIBGEN_H is set and gitbasename() from > compat/basename is used. > > Signed-off-by: Torsten Bögershausen > --- > My version of a hot fix > test-path-utils.c | 2 ++ > 1 file changed, 2 insertions(+) Thanks. Ramsay gave a nice analysis at http://thread.gmane.org/gmane.comp.version-control.git/283928 and Dscho already is working on it, IIUC, in a slightly different direction. > diff --git a/test-path-utils.c b/test-path-utils.c > index 4ab68ac..db0bb39 100644 > --- a/test-path-utils.c > +++ b/test-path-utils.c > @@ -142,7 +142,9 @@ static struct test_data dirname_data[] = { > { ".", "." }, > { "..", "." }, > { "/", "/" }, > +#if defined(NO_LIBGEN_H) > { "//", "//" }, > +#endif > #if defined(__CYGWIN__) && !defined(NO_LIBGEN_H) > { "///", "//" }, > { "////", "//" }, ``` ## Johannes Schindelin, 2016-01-14 07:00 Subject: Re: [PATCH/RFC] t0060: basename("//") is implementation defined Message-ID: URL: https://gitlist.dev/e/alpine.DEB.2.20.1601140758420.2964%40virtualbox In-Reply-To: <1452718561-32226-1-git-send-email-tboegi@web.de> ``` Hi Torsten, On Wed, 13 Jan 2016, tboegi@web.de wrote: > diff --git a/test-path-utils.c b/test-path-utils.c > index 4ab68ac..db0bb39 100644 > --- a/test-path-utils.c > +++ b/test-path-utils.c > @@ -142,7 +142,9 @@ static struct test_data dirname_data[] = { > { ".", "." }, > { "..", "." }, > { "/", "/" }, > +#if defined(NO_LIBGEN_H) > { "//", "//" }, > +#endif This would miss faulty implementations that return completely bogus outputs for the "//" input. Not very likely, I agree, but the entire intention of this regression test is to verify that Git's expectations are met (and to document those expectations). I hope you like the alternative patch I just sent out. Ciao, Dscho ``` ## Torsten Bögershausen, 2016-01-14 11:47 Subject: Re: [PATCH/RFC] t0060: basename("//") is implementation defined Message-ID: <56978ADF.5040401@web.de> URL: https://gitlist.dev/e/56978ADF.5040401%40web.de In-Reply-To: ``` On 01/14/2016 08:00 AM, Johannes Schindelin wrote: > Hi Torsten, > > On Wed, 13 Jan 2016, tboegi@web.de wrote: > >> diff --git a/test-path-utils.c b/test-path-utils.c >> index 4ab68ac..db0bb39 100644 >> --- a/test-path-utils.c >> +++ b/test-path-utils.c >> @@ -142,7 +142,9 @@ static struct test_data dirname_data[] = { >> { ".", "." }, >> { "..", "." }, >> { "/", "/" }, >> +#if defined(NO_LIBGEN_H) >> { "//", "//" }, >> +#endif > This would miss faulty implementations that return completely bogus > outputs for the "//" input. Not very likely, I agree, but the entire > intention of this regression test is to verify that Git's expectations are > met (and to document those expectations). > > I hope you like the alternative patch I just sent out. Yepp, looks good. Thanks for the work. ```