Re: chainlint test failing on Linux sparc64

On Mon, May 20, 2024 at 5:04 AM John Paul Adrian Glaubitz
[off-list ref] wrote:

Bisecting was successful and has lead me to this commit:

d00113ec3474a1652a73c11695c7e7b5182d80a7 is the first bad commit
Author: Eric Sunshine [off-list ref]
Date:   Thu Sep 1 00:29:46 2022 +0000
    t/Makefile: apply chainlint.pl to existing self-tests

    Now that chainlint.pl is functional, take advantage of the existing
    chainlint self-tests to validate its operation. (While at it, stop
    validating chainlint.sed against the self-tests since it will soon be
    retired.)

That's the point at which chainlint.sed was replaced with
chainlint.pl. It also leads to the same supposition expressed earlier
that the "actual" file seems to be empty for some reason, perhaps
because one of the commands which creates it is failing somehow. At
that point in time, the creation command for "actual" in t/Makefile
was:

    $(CHAINLINT) --emit-all '$(CHAINLINTTMP_SQ)'/tests | \
        grep -v '^[ ]*$$' >'$(CHAINLINTTMP_SQ)'/actual && \

which uses `grep` rather than the `sed` used in the present-day
t/Makefile. This may imply that Perl is somehow suspect or that
PERL_PATH or PERL_PATH_SQ in t/Makefile are incorrectly set, or that
t/chainlint.pl itself is not generating any output for some reason.

After running:

    cd t
    make check-chainlint

what is the output of:

    ls -l chainlinttmp

For me the output using the latest "master" branch is:

    % ls -l chainlinttmp
    -rw-------  1 me  me  14797 May 20 05:26 actual
    -rw-------  1 me  me  14797 May 20 05:26 expect
    -rw-------  1 me  me  15086 May 20 05:26 tests

My suspicion is that "actual" will have size 0 for you. If that's the
case, I'd suggest running the commands from the "check-chainlint"
target in t/Makefile manually one at a time to see if you can figure
out which is failing. For instance:

    % ./chainlint.pl --emit-all chainlinttmp/tests

If that produces output, then that's a good sign. Try capturing that
output and feeding it to the `sed` command:

    % ./chainlint.pl --emit-all chainlinttmp/tests >lint.out
    % sed -e 's/^[1-9][0-9]* //' lint.out

If that produces output, then something else is going wrong (such as
PERL_PATH or PERL_PATH_SQ being incorrect).