t/perf: fix test_export() failure with BSD `sed`
PATCH 1/1 t/perf: fix test_export() failure with BSD `sed`
test_perf() runs each test in its own subshell which makes it difficult to persist variables between tests. test_export() addresses this shortcoming by grabbing the values of specified variables after a test runs but before the subshell exits, and writes those values to a file which is loaded into the environment of subsequent tests.
To grab the values to be persisted, test_export() pipes the output of the shell's builtin set command through sed which plucks them out using a regular expression along the lines of s/^(var1|var2)/.../p. Unfortunately, though, this use of alternation is not portable. For instance, BSD-lineage sed (including macOS sed) does not support it in the default "basic regular expression" mode (BRE). It may be possible to enable "extended regular expression" mode (ERE) in some cases with sed -E, however, -E is neither portable nor part of POSIX.
Fortunately, alternation is unnecessary in this case and can easily be avoided, so replace it with a series of simple expressions such as s/^var1/.../p;s/^var2/.../p.
While at it, tighten the expressions so they match the variable names exactly rather than matching prefixes (i.e. use s/^var1=/.../p).
If the requirements of test_export() become more complex in the future, then an alternative would be to replace sed with perl which supports alternation on all platforms, however, the simple elimination of alternation via multiple sed expressions suffices for the present.
--- t/perf/perf-lib.sh | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)
t/perf/perf-lib.sh +7 -2
@@ -148,13 +148,18 @@ test_run_perf_ () { . '"$TEST_DIRECTORY"/test-lib-functions.sh' test_export () { [ $# != 0 ] || return 0 - test_export_="$test_export_\\|$1" + test_export_="$test_export_ $1" shift test_export "$@" } '"$1"' ret=$? -set | sed -n "s'"/'/'\\\\''/g"';s/^\\($test_export_\\)/export '"'&'"'/p" >test_vars +needles= +for v in $test_export_ +do + needles="$needles;s/^$v=/export $v=/p" +done +set | sed -n "s'"/'/'\\\\''/g"'$needles" >test_vars exit $ret' >&3 2>&4 eval_ret=$?
Signature · 1 line
2.30.0.rc0.297.gbcca948854
Quoted 3 lines · Eric Sunshine
Fortunately, alternation is unnecessary in this case and can easily be avoided, so replace it with a series of simple expressions such as
s/^var1/.../p;s/^var2/.../p.
Simple is good.
Quoted 2 lines
While at it, tighten the expressions so they match the variable names exactly rather than matching prefixes (i.e. use
s/^var1=/.../p).
Good eyes. That is quite good.
t/perf/perf-lib.sh test_run_perf_ () { · line 154
Quoted 9 lines
t/perf/perf-lib.sh +1
-1@@ -148,13 +148,18 @@ test_run_perf_ () { . '"$TEST_DIRECTORY"/test-lib-functions.sh' test_export () { [ $# != 0 ] || return 0 - test_export_="$test_export_\\|$1" + test_export_="$test_export_ $1" shift test_export "$@" }
This "recursion to consume $@ one by one, instead of looping" caught my eyes a bit, but the bug being fixed is not caused by it, so it is fine to let it pass. Given that the arguments to test_export are supposed to be all variable names (i.e. no funny characters anywhere in it), I think it could just be
test_export () {
test_export_="$*"
}though.
Oh, does anybody need to clear test_export_ to an empty string (or unset it), by the way?
t/perf/perf-lib.sh test_run_perf_ () { · line 164
Quoted 11 lines
t/perf/perf-lib.sh +6
-1'"$1"' ret=$? -set | sed -n "s'"/'/'\\\\''/g"';s/^\\($test_export_\\)/export '"'&'"'/p" >test_vars +needles= +for v in $test_export_ +do + needles="$needles;s/^$v=/export $v=/p" +done +set | sed -n "s'"/'/'\\\\''/g"'$needles" >test_vars exit $ret' >&3 2>&4 eval_ret=$?
Other than these, none of which were "this is wrong and needs to be fixed", I have no comments. The patch is quite readably done.
Will queue. Thanks.
Quoted 9 lines · Junio C Hamano
t/perf/perf-lib.sh test_run_perf_ () { · line 154
Quoted 7 lines · Eric Sunshine
test_export () { [ $# != 0 ] || return 0t/perf/perf-lib.sh +1
-1- test_export_="$test_export_\\|$1" + test_export_="$test_export_ $1" shift test_export "$@" }
Quoted 3 lines
This "recursion to consume $@ one by one, instead of looping" caught my eyes a bit, but the bug being fixed is not caused by it, so it is fine to let it pass.
The unusual recursion caught my eye as well.
Quoted 9 lines
Given that the arguments to
test_exportare
supposed to be all variable names (i.e. no funny characters anywhere in it), I think it could just betest_export () { test_export_="$*" }though.
That's almost good enough, but test_export() needs to accumulate the to-be-exported variables across multiple calls, so the non-recursive definition would likely be:
test_export () {
test_export_="$test_export_ $*"
}which would make a nice cleanup atop this portability-fix patch.
Quoted 2 lines
Oh, does anybody need to clear
test_export_to an empty string (or unset it), by the way?
Perhaps a test_unexport() might be handy in the distant future, but presently there is only a single call to test_export() in the entire suite, so it's probably not worth worrying about now.
Quoted 6 lines · Eric Sunshine
Quoted 2 lines
Oh, does anybody need to clear
test_export_to an empty string (or unset it), by the way?Perhaps a
test_unexport()might be handy in the distant future, but presently there is only a single call totest_export()in the entire suite, so it's probably not worth worrying about now.
I actually wonder if we could drop test_export entirely. I assume you mean the call in p0001. It is inside a test_expect_success block, where we don't need to do anything fancier than just "export". It is already running in the main script's environment, just like a normal test. If it were in a test_perf, then we would need to take special care to get it back into the main script.
There are some calls in p0000 like that, but they are really about testing test_export itself.
-Peff
Quoted 11 lines · Jeff King
Quoted 3 lines · Eric Sunshine
Perhaps a
test_unexport()might be handy in the distant future, but presently there is only a single call totest_export()in the entire suite, so it's probably not worth worrying about now.I actually wonder if we could drop
test_exportentirely. I assume you mean the call in p0001. It is inside atest_expect_successblock, where we don't need to do anything fancier than just "export". It is already running in the main script's environment, just like a normal test. If it were in atest_perf, then we would need to take special care to get it back into the main script.
Considering that test_export() hasn't seen much use since its introduction nine years ago and that the one and only existing call doesn't even need the special subprocess magic, retiring the function is certainly an option. On the other hand, aside from this one minor portability fix, it hasn't been a maintenance burden and may actually come in handy someday if people start writing more "perf" tests. So, I don't feel strongly one way or the other, though I lean somewhat toward keeping it around.
Quoted 13 lines · Eric Sunshine
Quoted 11 lines · Jeff King
Quoted 3 lines · Eric Sunshine
Perhaps a
test_unexport()might be handy in the distant future, but presently there is only a single call totest_export()in the entire suite, so it's probably not worth worrying about now.I actually wonder if we could drop
test_exportentirely. I assume you mean the call in p0001. It is inside atest_expect_successblock, where we don't need to do anything fancier than just "export". It is already running in the main script's environment, just like a normal test. If it were in atest_perf, then we would need to take special care to get it back into the main script.
Quoted 8 lines
Considering that
test_export()hasn't seen much use since its introduction nine years ago and that the one and only existing call doesn't even need the special subprocess magic, retiring the function is certainly an option. On the other hand, aside from this one minor portability fix, it hasn't been a maintenance burden and may actually come in handy someday if people start writing more "perf" tests. So, I don't feel strongly one way or the other, though I lean somewhat toward keeping it around.
That more or less matches my feeling. I just like deleting things. ;)
-Peff
Quoted 9 lines · Eric Sunshine
That's almost good enough, but
test_export()needs to accumulate the to-be-exported variables across multiple calls, so the non-recursive definition would likely be:test_export () { test_export_="$test_export_ $*" }which would make a nice cleanup atop this portability-fix patch.
A cleanup patch has been posted[1].
[PATCH 2/1] t/perf: avoid unnecessary test_export() recursion
In reply to Eric Sunshinetest_export() has been self-recursive since its inception even though a simple for-loop would have served just as well to append its arguments to the test_export_ variable separated by the pipe character "|". Recently test_export_ was changed instead to a space-separated list of tokens to be exported, an operation which can be accomplished via a single simple assignment, with no need for looping or recursion. Therefore, simplify the implementation.
While at it, take advantage of the fact that variable names to be exported are shell identifiers, thus won't be composed of special characters or whitespace, thus simple a $* can be used rather than magical "$@".
---
This is a follow-up to [1] as discussed in [2]. It applies cleanly atop 'es/perf-export-fix' and 'master'.
[1]: https://lore.kernel.org/git/[email protected]/
[2]: https://lore.kernel.org/git/CAPig+cR+4Wh4Sgk6UhUML4SHqaQsvYmw_77ih+oec2YmqQJCCg@mail.gmail.com/
t/perf/perf-lib.sh | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
t/perf/perf-lib.sh +1 -4
@@ -147,10 +147,7 @@ test_run_perf_ () { "$GTIME" -f "%E %U %S" -o test_time.$i "$SHELL" -c ' . '"$TEST_DIRECTORY"/test-lib-functions.sh' test_export () { - [ $# != 0 ] || return 0 - test_export_="$test_export_ $1" - shift - test_export "$@" + test_export_="$test_export_ $*" } '"$1"' ret=$?
Signature · 1 line
2.30.0.rc1.243.g5298b911bd