diff options
Diffstat (limited to 'contrib/bmake/unit-tests/job-output-null.mk')
| -rw-r--r-- | contrib/bmake/unit-tests/job-output-null.mk | 49 |
1 files changed, 36 insertions, 13 deletions
diff --git a/contrib/bmake/unit-tests/job-output-null.mk b/contrib/bmake/unit-tests/job-output-null.mk index 7620bdf6a7ba..04786dba4096 100644 --- a/contrib/bmake/unit-tests/job-output-null.mk +++ b/contrib/bmake/unit-tests/job-output-null.mk @@ -1,32 +1,55 @@ -# $NetBSD: job-output-null.mk,v 1.1 2021/04/15 19:02:29 rillig Exp $ +# $NetBSD: job-output-null.mk,v 1.4 2022/09/03 08:03:27 rillig Exp $ # # Test how null bytes in the output of a command are handled. Make processes # them using null-terminated strings, which may cut off some of the output. # -# As of 2021-04-15, make handles null bytes from the child process -# inconsistently. It's an edge case though since typically the child -# processes output text. +# Before job.c 1.454 from 2022-09-03, make handled null bytes in the output +# from the child process inconsistently. It's an edge case though since +# typically the child processes output text. + +# Note: The printf commands used in this test must only use a single format +# string, without parameters. This is because it is implementation-dependent +# how many times the command 'printf "fmt%s" "" "" ""' calls write(2). +# +# NetBSD /bin/sh 1 x write("fmtfmtfmt") +# Dash 1 x write("fmtfmtfmt") +# NetBSD /bin/ksh 3 x write("fmt") (via /bin/printf) +# Bash 5 3 x write("fmt") +# +# In the latter case the output may arrive in 1 to 3 parts, depending on the +# exact timing, which in this test makes a crucial difference since before +# job.c 1.454 from 2022-09-03, the outcome of the test depended on whether +# there was a '\n' in each of the blocks from the output. Depending on the +# exact timing, the output of that test varied, its possible values were '2a', +# '2a 2b', '2a 2c', '2a 2b 2c'. .MAKEFLAGS: -j1 # force jobs mode all: .PHONY - # The null byte from the command output is kept as-is. - # See CollectOutput, which looks like it intended to replace these - # null bytes with simple spaces. - @printf 'hello\0world%s\n' '' + # The null byte from the command output is replaced with a single + # space by CollectOutput. + @printf '1\0trailing\n' + # expect: 1 trailing # Give the parent process a chance to see the above output, but not # yet the output from the next printf command. @sleep 1 - # All null bytes from the command output are kept as-is. - @printf 'hello\0world%s\n' '' '' '' '' '' '' + # Each null byte from the command output is replaced with a single + # space. + @printf '2a\0trailing\n''2b\0trailing\n''2c\0trailing\n' + # expect: 2a trailing + # expect: 2b trailing + # expect: 2c trailing @sleep 1 - # The null bytes are replaced with spaces since they are not followed - # by a newline. + # Each null byte from the command output is replaced with a single + # space. Because there is no trailing newline in the output, these + # null bytes were replaced with spaces even before job.c 1.454 from + # 2022-09-03, unlike in the cases above. # # The three null bytes in a row test whether this output is # compressed to a single space like in DebugFailedTarget. It isn't. - @printf 'hello\0world\0without\0\0\0newline%s' ', ' ', ' '.' + @printf '3a\0without\0\0\0newline, 3b\0without\0\0\0newline.' + # expect: 3a without newline, 3b without newline. |