On Tue, Mar 09, 2021 at 09:57:03PM +0100, Ævar Arnfjörð Bjarmason wrote:
> >> diff --git a/t/README b/t/README
> >> index 2cc8cbc7185..f09d94e754e 100644
> >> --- a/t/README
> >> +++ b/t/README
> >> @@ -157,10 +157,13 @@ appropriately before running "make". Short options can be bundled, i.e.
> >>
> >> -V::
> >> --verbose-log::
> >> - Write verbose output to the same logfile as `--tee`, but do
> >> - _not_ write it to stdout. Unlike `--tee --verbose`, this option
> >> + An alias for `--verbose --tee`. This option
> >> is safe to use when stdout is being consumed by a TAP parser
> >> - like `prove`. Implies `--tee` and `--verbose`.
> >> + like `prove`.
> >> + Historically this option was different from `--verbose --tee`
> >> + and would not write any verbose output to stdout to ensure the
> >> + TAP-correctness of the output. The TAP-correctness of the
> >> + output is now sanity checked by the test library,
> >
> > Not everyone is using a TAP harness to run the tests, and, therefore,
> > '--verbose-log' should not spew out verbose output to the terminal.
>
> IOW even though --verbose-log was meant as a hack to make prove happy,
> you've since come to like the "verbose in log, but not stdout" mode and
> want that kept?
Yes, '--verbose-log' proved to be really convenient, even if it was
meant to solve a different issue.
> Yes, this patch takes that mode away.
>
> Yes, I can change it.
>
> Would your use-case for this be satisfied by having prove(1) just emit
> different output for you in this scenario, so you'd need to invoke this
> as something like:
>
> prove <test> :: --verbose --tee # or --verbose-log
I use prove to run the test suite, but I don't and won't use prove to
run a single test.
The behavior of './t1234-foo.sh -V' with or without '-x' should not
change without _very_ convincing reasons. "We now output valid TAP
even with --verbose, so we don't need it for the TAP harness" is
definitely not convincing.
> Becuse the advantage of this series is that that sort of thing becomes
> really trivial without everything needing to be hardcoded in
> test-lib.sh, observe (this is with my series):
>
>
> 0 $ PERL5LIB=. prove -v --formatter=SZEDERVerboseLog ./t0201-gettext-fallbacks.sh :: --verbose-log
Well, this doesn't look trivial at all, does it? In fact, I consider
this unusably convoluted.
> # lib-gettext: No is_IS UTF-8 locale available
> # lib-gettext: No is_IS ISO-8859-1 locale available
> ok 1 - sanity: $GIT_INTERNAL_GETTEXT_SH_SCHEME is set (to fallthrough)
> ok 2 - sanity: $GIT_INTERNAL_GETTEXT_TEST_FALLBACKS is set
> ok 3 - sanity: $GIT_INTERNAL_GETTEXT_SH_SCHEME" is fallthrough
> ok 4 - gettext: our gettext() fallback has pass-through semantics
> ok 5 - eval_gettext: our eval_gettext() fallback has pass-through semantics
> ok 6 - eval_gettext: our eval_gettext() fallback can interpolate variables
> ok 7 - eval_gettext: our eval_gettext() fallback can interpolate variables with spaces
> ok 8 - eval_gettext: our eval_gettext() fallback can interpolate variables with spaces and quotes
> # passed all 8 test(s)
> 1..8
> ok
> All tests successful.
> Files=1, Tests=8, 1 wallclock secs ( 0.01 usr 0.01 sys + 0.07 cusr 0.01 csys = 0.10 CPU)
> Result: PASS
> $ wc -l test-results/t0201-gettext-fallbacks.out
> 75 test-results/t0201-gettext-fallbacks.out
>
> All without any patching on top to the test-lib.sh, just:
>
> $ cat SZEDERVerboseLog.pm
> package SZEDERVerboseLog::Session;
> use base 'TAP::Formatter::Console::Session';
>
> sub result {
> my ($self, $result) = @_;
> return if $result->is_unknown;
> return $self->SUPER::result($result);
> }
>
> package SZEDERVerboseLog;
> use strict;
> use warnings;
> use base 'TAP::Formatter::Console';
>
> sub open_test {
> my ($self, $test, $parser) = @_;
> my $session = SZEDERVerboseLog::Session->new( {
> name => $test,
> formatter => $self,
> parser => $parser,
> } );
> return $session;
> }
>
> 1;
>
> The "is_unknown" is everything that's not TAP syntax.
>
