Hi Paul Ok mystery solved :). Once I have this fixed, what should I do? Push the new tarball with the same version (1.0.1-1) or should I bump something in the version? I'd expect that perhaps that -1 should become a -2? If so, what's the correct way of doing this?
Cheers, Gabriele On Wed, 3 Jun 2020 at 16:36, Paul Gevers <elb...@debian.org> wrote: > > Hi Gabriele, > > On 02-06-2020 23:37, Gabriele wrote: > > Many thanks for your reply. I have had a look at the logs linked on this > > page > > > > https://ci.debian.net/packages/a/austin/testing/amd64/ > > > > The only version that passes is v1.0.0 and by looking at the logs of > > v0.7.0 and v1.0.1, which fail, it's a miracle that v1.0.0 even passes. > > Indeed v0.7.0 and v1.0.1 fail for the very same reason: the binary > > used for the tests, src/austin, is simply not there. Why it is there > > for the v1.0.0 I don't know, so it looks like the problem is with > > v1.0.0 paradoxically. > > It's funny, the first tries of 1.0.0 also failed. And I believe they > they only starting passing when python3.7 was not the default Python3 > anymore. > > > This is the diff inside the debian/ folder between v1.0.0 and v1.0.1 > > (TLDR: only debian/austin.1, debian/changelog and debian/copyright > > have changed) > > Instead, I diffed the source and this struck my eye: > > diff -Nru austin-1.0.0/test/test_fork.bats austin-1.0.1/test/test_fork.bats > --- austin-1.0.0/test/test_fork.bats 2019-10-19 10:37:23.000000000 +0000 > +++ austin-1.0.1/test/test_fork.bats 2020-02-21 19:27:02.000000000 +0000 > @@ -56,6 +56,12 @@ > then > skip "Test failed but marked as 'Ignore'" > else > + echo > + echo "Collected Output" > + echo "================" > + echo > + echo "$output" > + echo > false > fi > } > @@ -109,6 +115,6 @@ > invoke_austin "3.7" > } > > -# @test "Test Austin with Python 3.8" { > -# invoke_austin "3.8" > -# } > +@test "Test Austin with Python 3.8" { > + invoke_austin "3.8" > +} > > So, with 1.0.0 your tests were passing because all tests were skipped, > and only with 1.0.1 your tests started testing the code again and failed > because the required binary couldn't be found. > > > > Hence, to the best of my knowledge, there are no changes in the > > debian/ area that would cause the binary in src/ to be there unless it > > accidentally ended up, say, in the source tarball. > > I think I've showed an alternative explanation. > > > My next question to you is then: where is the binary supposed to be > > found during autopkgtest? Can I assume it will be on the PATH during > > testing, so that I can invoke it simply with "austin"? Or do I need to > > specify a precise path? > > The testbed is a clean Debian installation, created with debbootstrap > and only your test dependencies installed. Everything is in it's regular > location, so if austin is on the regular path for users, it on the > regular path for the debci user in the testbed. Not specifying the > precise path makes sure your testing that it's on the path for your > users too, so better without path. > > Paul > -- "Egli è scritto in lingua matematica, e i caratteri son triangoli, cerchi, ed altre figure geometriche, senza i quali mezzi è impossibile a intenderne umanamente parola; senza questi è un aggirarsi vanamente per un oscuro laberinto." -- G. Galilei, Il saggiatore.
