003_extrafiles.pl test fails on Windows with the newer Perl versions - Mailing list pgsql-hackers

From Nazir Bilal Yavuz
Subject 003_extrafiles.pl test fails on Windows with the newer Perl versions
Date
Msg-id CAN55FZ2XW1B5NP11Vy-y9O2BmMnLFq8grCsHZr_AovcL4QCgdg@mail.gmail.com
Whole thread Raw
Responses Re: 003_extrafiles.pl test fails on Windows with the newer Perl versions
Re: 003_extrafiles.pl test fails on Windows with the newer Perl versions
List pgsql-hackers
Hi,

I was trying to install newer Perl versions to Windows CI images and
found that 003_extrafiles.pl test fails on Windows with:

(0.183s) not ok 2 - file lists match
(0.000s) #   Failed test 'file lists match'
#   at C:/cirrus/src/bin/pg_rewind/t/003_extrafiles.pl line 81.
(0.000s) #     Structures begin differing at:
#          $got->[0] =
'C:/cirrus/build/testrun/pg_rewind/003_extrafiles/data/t_003_extrafiles_primary_local_data/pgdata/tst_both_dir'
#     $expected->[0] =
'C:\cirrus\build/testrun/pg_rewind/003_extrafiles\data/t_003_extrafiles_primary_local_data/pgdata/tst_both_dir'

(0.263s) not ok 5 - file lists match
(0.000s) #   Failed test 'file lists match'
#   at C:/cirrus/src/bin/pg_rewind/t/003_extrafiles.pl line 81.
(0.000s) #     Structures begin differing at:
#          $got->[0] =
'C:/cirrus/build/testrun/pg_rewind/003_extrafiles/data/t_003_extrafiles_primary_remote_data/pgdata/tst_both_dir'
#     $expected->[0] =
'C:\cirrus\build/testrun/pg_rewind/003_extrafiles\data/t_003_extrafiles_primary_remote_data/pgdata/tst_both_dir'

It looks like File::Find converts backslashes to slashes in the newer
Perl versions. I tried to find the related commit and found this:
https://github.com/Perl/perl5/commit/414f14df98cb1c9a20f92c5c54948b67c09f072d

To solve this, I did the same conversion for Windows before comparing
the paths. And to support all Perl versions, I decided to always
convert them on Windows regardless of the Perl's version. The fix is
attached.

I looked at other File::Find appearances in the code but they do not
compare the paths. So, I do not think there is any need to fix them.

Any kind of feedback would be appreciated.

--
Regards,
Nazir Bilal Yavuz
Microsoft

Attachment

pgsql-hackers by date:

Previous
From: Alexander Kuzmenkov
Date:
Subject: Re: Incorrect cost for MergeAppend
Next
From: Matthias van de Meent
Date:
Subject: Re: Reducing output size of nodeToString