mbarbon/perl-remote-debugging-client

BBC: tests have started to fail with bleadperl 5.37.9

Opened this issue · 3 comments

andk commented

Sample fail report: http://www.cpantesters.org/cpan/report/687b3dbc-b384-11ed-abd6-f6f02f9d4dbf

Git bisect leads me to:

commit 21938ae5a278f0b48443561333c47714921e7566 (HEAD)
Author: Yves Orton <demerphq@gmail.com>
Date:   Sun Sep 25 15:09:52 2022 +0200

    perl.h, pp_ctl.c - switch to standard way of terminating compilation

The commit is also known as v5.37.8-202-g21938ae5a2 ; Yves Orten is also known as @demerphq ;

HI @mbarbon hope you are well. Long time no talk.

I havent quite got to the bottom of how to best fix this. The main point is that eval compile errors now behave consistently, and will always include the verbiage "Execution of ... aborted due to compilation errors", where previously it would or would not do so depending on the nature and count of errors in the code being compiled. Eg:

syntax error at (eval 18)[blib/lib/dbgp-helper/perl5db.pl:192] line 1, at EOF
Execution of (eval 18)[blib/lib/dbgp-helper/perl5db.pl:192] aborted due to compilation errors.

I havent found the code that scrubs the (eval 18)... from the old message, based on similar bugs ive fixed i guess that is where you should scrub this out too. (Assuming it exists).

I did get to the bottom of something else however, the Makefile.PL should have this applied:

diff --git a/Makefile.PL b/Makefile.PL
index 964aa3a..5e8503d 100644
--- a/Makefile.PL
+++ b/Makefile.PL
@@ -67,7 +67,7 @@ WriteMakefile(
     },
     clean             => { FILES => 'DB/DbgrXS.pm' },
     $do_xs ? (
-        XS                => { 'perl5db.xs' => undef },
+        XS                => { 'perl5db.xs' => 'perl5db.c' },
         OBJECT            => 'perl5db.o',
     ) : (
         XS                => { },

which will silence a bunch of use of uninitialized variable warnings when executing Makefile.PL

I will continue to poke into this to see if I can fix it. But i think it might be that a different module entirely needs to be fixed. I am not entirely sure yet.

@andk it happens that @mbarbon and I know each other.

@mbarbon I have pushed a PR to fix this. It does not include a version bump.