Martin, The patch was always already calling the 'cat' binary, it just changed the argv[0] to also check that 'feature' of setting the process name to something else. But, as Sebastian says, that breaks with 'cat' and is not really relevant to be tested here. So the patch is really addressing this issue, no question about it.
-Christian On 1/6/21 3:28 AM, Martin Schanzenbach wrote: > Hello Sebastian, > > Thanks! > I am not sure your patch addresses the issue. Or maybe I do not > understand it. > If you replace the binary that is called through > GNUNET_OS_start_process from the testfile to "cat", then the test is no > longer doing what it is supposed to do, right? > > Can you elaborate a bit more on what is happening when the test fails > and how you patch addresses the problem? I understand that "cat" in > busybox may just be a call to binary from a specific name but how does > that relate to the test? > > BR > > On Tue, 2021-01-05 at 17:43 -0300, Sebastian Javier Marchano wrote: >> Hello! >> >> I was playing around with gnunet and alpine and it seems that this >> test (os start process) breaks because alpine uses the same binary >> for some commands (/bin/busybox) and relies on the filename for the >> command name. >> >> Best regards >> >> diff --git a/src/util/test_os_start_process.c >> b/src/util/test_os_start_process.c >> index ff5021d03..7e5cde1a9 100644 >> --- a/src/util/test_os_start_process.c >> +++ b/src/util/test_os_start_process.c >> @@ -138,7 +138,7 @@ run_task (void *cls) >> GNUNET_OS_start_process (GNUNET_OS_INHERIT_STD_ERR, >> hello_pipe_stdin, hello_pipe_stdout, >> NULL, >> fn, >> - "test_gnunet_echo_hello", "-", NULL); >> + "cat", "-", NULL); >> GNUNET_free (fn); >> >> /* Close the write end of the read pipe */ >> >> >> >> -- >> Sebastian >
signature.asc
Description: OpenPGP digital signature
