It worked on my systems, but then I laid down the framework the way travis does and it fails like travis
in a subdir of your source tree: git clone --depth=1 https://github.com/apache/httpd-tests.git test/perl-framework On Sat, Sep 18, 2021 at 1:10 PM Christophe JAILLET <christophe.jail...@wanadoo.fr> wrote: > > Le 17/09/2021 à 16:38, Travis CI a écrit : > > apache > > > > / > > > > httpd > > > > <https://app.travis-ci.com/github/apache/httpd?utm_medium=notification&utm_source=email> > > > > > > branch icontrunk <https://github.com/apache/httpd/tree/trunk> > > > > build has failed > > Build #1937 was broken > > <https://app.travis-ci.com/github/apache/httpd/builds/237906294?utm_medium=notification&utm_source=email> > > arrow to build time > > clock icon11 mins and 56 secs > > > > Hi, > > The error is related to some new tests I've added (r1893389) to improve > our test-coverage. > > The easiest would be to remove these (bogus ?) new tests. > They are likely due to my misunderstanding of how these directives work. > > However, the tests work just fine on my config, so I don't really see > why it fails on travis. > > Some AH00128 are logged, so the 'default_handler' from core.c is called. > I don't see such errors in my logs. I've got only 3, which are expected. > Two because .xyz and .xyz22 files have no handler, one because of a GET > without any query arguments (e.g., foo.html?hi). > > The failing tests are: > t/modules/actions.t ................. 1/20 > # Failed test 9 in t/modules/actions.t at line 32 fail #5 > # Failed test 10 in t/modules/actions.t at line 33 fail #2 > # Failed test 13 in t/modules/actions.t at line 52 > # Failed test 14 in t/modules/actions.t at line 53 > # Failed test 16 in t/modules/actions.t at line 43 fail #2 > # Failed test 17 in t/modules/actions.t at line 44 > # Failed test 18 in t/modules/actions.t at line 52 fail #2 > # Failed test 19 in t/modules/actions.t at line 53 fail #2 > t/modules/actions.t ................. Failed 8/20 subtests > > Test 9 and 10 try to access a non-existent file on purpose. In this > case, the 'action' is configured as 'virtual' and the non-existence of > the file should not be a problem. > > The other tests are related to 'Script'. In this case, my understanding > is that we could only end to the 'default_handler' if we are looping. > In such a case, I don't see why it would behave differently on my config > and on travis. > > :\ > > CJ -- Eric Covener cove...@gmail.com
