search

Perl / perl5

🐪 The Perl programming language
https://dev.perl.org/perl5/
Other
1.91k stars 542 forks source link

BBC: Blead breaks GRIAN/Storable-AMF-1.23.tar.gz #19462

Closed andk closed 2 years ago

andk commented 2 years ago

Description

Commit v5.35.8-110-gbb5bc97fde/da1c (https://github.com/perl/perl5/commit/bb5bc9) stopped also GRIAN/Storable-AMF-1.23.tar.gz from succeeding with the test suite. We had this commit already in https://github.com/Perl/perl5/issues/19450 ; that issue has been closed, but the fix provided over there does not help for Storable::AMF 1.23 ; you can already find sample fail reports on http://matrix.cpantesters.org/?dist=Storable-AMF+1.23

Steps to Reproduce

cpan -i GRIAN/Storable-AMF-1.23.tar.gz

Expected behavior

Should successfully test and install the module

Perl configuration

# perl -V output goes here
Summary of my perl5 (revision 5 version 35 subversion 10) configuration:
  Commit id: 714a0a851bee9b5ed8d5e0f66865bf3bb1d2bbc5
  Platform:
    osname=linux
    osvers=5.15.0-2-amd64
    archname=x86_64-linux-ld
    uname='linux k93msid 5.15.0-2-amd64 #1 smp debian 5.15.5-2 (2021-12-18) x86_64 gnulinux '
    config_args='-Dprefix=/home/sand/src/perl/repoperls/installed-perls/host/k93msid/v5.35.9-18-g714a0a851b/ea6b -Dmyhostname=k93msid -Dinstallusrbinperl=n -Uversiononly -Dusedevel -des -Ui_db -Dlibswanted=cl pthread socket inet nsl gdbm dbm malloc dl ld sun m crypt sec util c cposix posix ucb BSD gdbm_compat -Uuseithreads -Duselongdouble -DEBUGGING=none'
    hint=recommended
    useposix=true
    d_sigaction=define
    useithreads=undef
    usemultiplicity=undef
    use64bitint=define
    use64bitall=define
    uselongdouble=define
    usemymalloc=n
    default_inc_excludes_dot=define
  Compiler:
    cc='cc'
    ccflags ='-fwrapv -fno-strict-aliasing -pipe -fstack-protector-strong -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -D_FORTIFY_SOURCE=2'
    optimize='-O2'
    cppflags='-fwrapv -fno-strict-aliasing -pipe -fstack-protector-strong -I/usr/local/include'
    ccversion=''
    gccversion='11.2.0'
    gccosandvers=''
    intsize=4
    longsize=8
    ptrsize=8
    doublesize=8
    byteorder=12345678
    doublekind=3
    d_longlong=define
    longlongsize=8
    d_longdbl=define
    longdblsize=16
    longdblkind=3
    ivtype='long'
    ivsize=8
    nvtype='long double'
    nvsize=16
    Off_t='off_t'
    lseeksize=8
    alignbytes=16
    prototype=define
  Linker and Libraries:
    ld='cc'
    ldflags =' -fstack-protector-strong -L/usr/local/lib'
    libpth=/usr/local/lib /usr/lib/x86_64-linux-gnu /usr/lib /lib/x86_64-linux-gnu /lib
    libs=-lpthread -lnsl -ldl -lm -lcrypt -lutil -lc
    perllibs=-lpthread -lnsl -ldl -lm -lcrypt -lutil -lc
    libc=libc-2.33.so
    so=so
    useshrplib=false
    libperl=libperl.a
    gnulibc_version='2.33'
  Dynamic Linking:
    dlsrc=dl_dlopen.xs
    dlext=so
    d_dlsymun=undef
    ccdlflags='-Wl,-E'
    cccdlflags='-fPIC'
    lddlflags='-shared -O2 -L/usr/local/lib -fstack-protector-strong'

Characteristics of this binary (from libperl): 
  Compile-time options:
    HAS_TIMES
    PERLIO_LAYERS
    PERL_COPY_ON_WRITE
    PERL_DONT_CREATE_GVSV
    PERL_MALLOC_WRAP
    PERL_OP_PARENT
    PERL_PRESERVE_IVUV
    PERL_USE_DEVEL
    USE_64_BIT_ALL
    USE_64_BIT_INT
    USE_LARGE_FILES
    USE_LOCALE
    USE_LOCALE_COLLATE
    USE_LOCALE_CTYPE
    USE_LOCALE_NUMERIC
    USE_LOCALE_TIME
    USE_LONG_DOUBLE
    USE_PERLIO
    USE_PERL_ATOF
  Built under linux
  Compiled at Feb 24 2022 18:16:53
  @INC:
    /home/sand/src/perl/repoperls/installed-perls/host/k93msid/v5.35.9-18-g714a0a851b/ea6b/lib/site_perl/5.35.10/x86_64-linux-ld
    /home/sand/src/perl/repoperls/installed-perls/host/k93msid/v5.35.9-18-g714a0a851b/ea6b/lib/site_perl/5.35.10
    /home/sand/src/perl/repoperls/installed-perls/host/k93msid/v5.35.9-18-g714a0a851b/ea6b/lib/5.35.10/x86_64-linux-ld
    /home/sand/src/perl/repoperls/installed-perls/host/k93msid/v5.35.9-18-g714a0a851b/ea6b/lib/5.35.10
andk commented 2 years ago

Also affected: BRUMMETT/Devel-Chitin-0.18.tar.gz

jkeenan commented 2 years ago

Description

Commit v5.35.8-110-gbb5bc97fde/da1c (bb5bc9) stopped also GRIAN/Storable-AMF-1.23.tar.gz from succeeding with the test suite. We had this commit already in #19450 ; that issue has been closed, but the fix provided over there does not help for Storable::AMF 1.23 ; you can already find sample fail reports on http://matrix.cpantesters.org/?dist=Storable-AMF+1.23

[snip]

Failures confirmed with the following recent build on FreeBSD-12:

$ bleadperl -v | head -2 | tail -1
This is perl 5, version 35, subversion 10 (v5.35.10 (v5.35.9-7-g8432359dfe)) built for amd64-freebsd-thread-multi

$ bleadprove -vb t/64-prefer-number.t t/65-prefer-number3.t 
t/64-prefer-number.t ... 
1..44
ok 1 - int constant
ok 2 - double constant
[snip]
ok 12 - str var not changed
not ok 13 - Int converted  is   a string
ok 14 - Int 0+converted     is double
ok 15 - Int 0.0+converted   is double
ok 16 - Int converted again is double
not ok 17 - Int++ converted  is   a string

#   Failed test 'Int converted  is   a string'
#   at t/64-prefer-number.t line 37.

#   Failed test 'Int++ converted  is   a string'
#   at t/64-prefer-number.t line 47.
ok 18 - Int++ 0+converted     is double
ok 19 - Int++ 0.0+converted   is double
[snip]
ok 44 - Str 1.0 ''.converted       is str
# Looks like you failed 2 tests of 44.
Dubious, test returned 2 (wstat 512, 0x200)
Failed 2/44 subtests 
t/65-prefer-number3.t .. 
1..44
ok 1 - int constant
ok 2 - double constant
ok 3 - str constant
[snip]
ok 12 - str var not changed
not ok 13 - Int converted  is   a string

#   Failed test 'Int converted  is   a string'
#   at t/65-prefer-number3.t line 37.
ok 14 - Int 0+converted     is double
ok 15 - Int 0.0+converted   is double
ok 16 - Int converted again is double
not ok 17 - Int++ converted  is   a string

#   Failed test 'Int++ converted  is   a string'
#   at t/65-prefer-number3.t line 47.
ok 18 - Int++ 0+converted     is double
ok 19 - Int++ 0.0+converted   is double
ok 20 - Int++ converted again is double
[snip]
ok 44 - Str 1.0 ''.converted       is str
# Looks like you failed 2 tests of 44.
Dubious, test returned 2 (wstat 512, 0x200)
Failed 2/44 subtests 

Test Summary Report
-------------------
t/64-prefer-number.t (Wstat: 512 Tests: 44 Failed: 2)
  Failed tests:  13, 17
  Non-zero exit status: 2
t/65-prefer-number3.t (Wstat: 512 Tests: 44 Failed: 2)
  Failed tests:  13, 17
  Non-zero exit status: 2
Files=2, Tests=88,  0 wallclock secs ( 0.05 usr  0.02 sys +  0.20 cusr  0.04 csys =  0.30 CPU)
Result: FAIL
jkeenan commented 2 years ago

Also affected: BRUMMETT/Devel-Chitin-0.18.tar.gz

Partial output:

$ bleadprove -vb t/20-optree-deparse.t
t/20-optree-deparse.t .. 
# Seeded srand with seed '20220225' from local date.
1..29
# Subtest: t/20-optree-deparse/01-assignment
    1..11
    ok 1 - t/20-optree-deparse/01-assignment/array_ref_assignment
    ok 2 - t/20-optree-deparse/01-assignment/array_ref_slice_assignment
    ok 3 - t/20-optree-deparse/01-assignment/hasf_ref_slice_assignment
    ok 4 - t/20-optree-deparse/01-assignment/hash_assignment
    ok 5 - t/20-optree-deparse/01-assignment/hash_ref_assignment
    ok 6 - t/20-optree-deparse/01-assignment/hash_slice_assignment
    ok 7 - t/20-optree-deparse/01-assignment/list-assignment
    ok 8 - t/20-optree-deparse/01-assignment/list_index_assignment
    ok 9 - t/20-optree-deparse/01-assignment/list_slice
    ok 10 - t/20-optree-deparse/01-assignment/list_slice_assignment
    ok 11 - t/20-optree-deparse/01-assignment/scalar_ref_assignment
ok 1 - Subtest: t/20-optree-deparse/01-assignment
[snip]

# Subtest: t/20-optree-deparse/12-i-o
    1..29
    ok 1 - t/20-optree-deparse/12-i-o/binmode
    ok 2 - t/20-optree-deparse/12-i-o/close
    ok 3 - t/20-optree-deparse/12-i-o/closedir
    ok 4 - t/20-optree-deparse/12-i-o/dbmclose
    ok 5 - t/20-optree-deparse/12-i-o/dbmopen
    ok 6 - t/20-optree-deparse/12-i-o/die
    ok 7 - t/20-optree-deparse/12-i-o/eof
    ok 8 - t/20-optree-deparse/12-i-o/fileno
    ok 9 - t/20-optree-deparse/12-i-o/flock
    ok 10 - t/20-optree-deparse/12-i-o/getc
    ok 11 - t/20-optree-deparse/12-i-o/print
    ok 12 - t/20-optree-deparse/12-i-o/printf
    ok 13 - t/20-optree-deparse/12-i-o/read
    ok 14 - t/20-optree-deparse/12-i-o/readdir
    ok 15 - t/20-optree-deparse/12-i-o/readline
    ok 16 - t/20-optree-deparse/12-i-o/rewinddir
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
    not ok 17 - t/20-optree-deparse/12-i-o/seek

    # Failed test 't/20-optree-deparse/12-i-o/seek'
    # at t/20-optree-deparse.t line 90.
    # +-----------------------+----+-------------------------------+
    # | GOT                   | OP | CHECK                         |
    # +-----------------------+----+-------------------------------+
    # | $a = seek(F, 10, );\n | eq | $a = seek(F, 10, SEEK_CUR);\n |
    # | $fh;\n                |    | $fh;\n                        |
    # | seek($fh, -10, );\n   |    | seek($fh, -10, SEEK_END);\n   |
    # | seek(*$fh, 0, )\n     |    | seek(*$fh, 0, SEEK_SET)\n     |
    # +-----------------------+----+-------------------------------+
    # Showing whitespace:
    # >>$a = seek(F, 10, );<<
    # >>$fh;<<
    # >>seek($fh, -10, );<<
    # >>seek(*$fh, 0, )<<
    # $@: 
    # Tree:
UNOP leavesub (KIDS)  BARE, SLICE 0x802444300
  LISTOP lineseq (KIDS, PARENS)   0x802444dc8
    COP nextstate (WANT_LIST, WANT_VOID) (eval 177):2  0x802445060
    BINOP sassign (KIDS, STACKED, WANT_LIST, WANT_VOID)   0x8024453a8
      LISTOP seek (KIDS, WANT_LIST, WANT_SCALAR)   0x8024454b0
        OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x8024457c8
        UNOP rv2gv (KIDS, SPECIAL, WANT_LIST, WANT_SCALAR)  BARE, SLICE 0x802445af8
          UNOP entersub (KIDS, STACKED, WANT_LIST, WANT_SCALAR)   0x802445bb8
            BINOP null (ex-pp_list) (KIDS, WANT_LIST, WANT_SCALAR)   0x802445ed0
              OpTree pushmark (WANT_LIST, WANT_SCALAR)   0x802447228
              UNOP null (ex-pp_rv2cv) (KIDS, WANT_LIST, WANT_SCALAR)  KVSLICE 0x802447558
                PADOP gv (WANT_LIST, WANT_SCALAR)   0x802447630
        SVOP const (WANT_LIST, WANT_SCALAR) 10  0x802447978
        SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x802447b40
      OpTree padsv (MOD, REF, SPECIAL, WANT_LIST, WANT_SCALAR) $a  0x802447c18
    COP nextstate (WANT_LIST, WANT_VOID) (eval 177):3  0x802447d20
    OpTree padsv (MOD, WANT_LIST, WANT_VOID) $fh  0x802447e28
    COP nextstate (WANT_LIST, WANT_VOID) (eval 177):4  0x802447f00
    LISTOP seek (KIDS, WANT_LIST, WANT_VOID)   0x802447fd8
      OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x8024490d8
      UNOP rv2gv (KIDS, SPECIAL, WANT_LIST, WANT_SCALAR)  BARE, SLICE 0x802449180
        OpTree padsv (WANT_LIST, WANT_SCALAR) $fh  0x802449240
      SVOP const (WANT_LIST, WANT_SCALAR) -10  0x802449300
      SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x802449408
    COP nextstate (WANT_LIST, WANT_VOID) (eval 177):5  0x8024494e0
    LISTOP seek (KIDS, WANT_LIST, WANT_SCALAR)   0x8024495e8
      OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x8024496c0
      UNOP rv2gv (KIDS, REF, WANT_LIST, WANT_SCALAR)  BARE, SLICE 0x802449768
        OpTree padsv (WANT_LIST, WANT_SCALAR) $fh  0x802449828
      SVOP const (WANT_LIST, WANT_SCALAR) 0  0x8024498e8
      SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x8024499f0
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.

    ok 18 - t/20-optree-deparse/12-i-o/seekdir
    ok 19 - t/20-optree-deparse/12-i-o/select
    ok 20 - t/20-optree-deparse/12-i-o/select-syscall
    ok 21 - t/20-optree-deparse/12-i-o/syscall
    ok 22 - t/20-optree-deparse/12-i-o/sysread
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
    not ok 23 - t/20-optree-deparse/12-i-o/sysseek

    # Failed test 't/20-optree-deparse/12-i-o/sysseek'
    # at t/20-optree-deparse.t line 90.
    # +--------------------------+----+----------------------------------+
    # | GOT                      | OP | CHECK                            |
    # +--------------------------+----+----------------------------------+
    # | $a = sysseek(F, 10, );\n | eq | $a = sysseek(F, 10, SEEK_CUR);\n |
    # | $fh;\n                   |    | $fh;\n                           |
    # | sysseek($fh, -10, );\n   |    | sysseek($fh, -10, SEEK_END);\n   |
    # | sysseek(*$fh, 0, )\n     |    | sysseek(*$fh, 0, SEEK_SET)\n     |
    # +--------------------------+----+----------------------------------+
    # Showing whitespace:
    # >>$a = sysseek(F, 10, );<<
    # >>$fh;<<
    # >>sysseek($fh, -10, );<<
    # >>sysseek(*$fh, 0, )<<
    # $@: 
    # Tree:
UNOP leavesub (KIDS)  BARE, SLICE 0x8024958a0
  LISTOP lineseq (KIDS, PARENS)   0x802495eb8
    COP nextstate (WANT_LIST, WANT_VOID) (eval 183):1  0x802496498
    BINOP sassign (KIDS, STACKED, WANT_LIST, WANT_VOID)   0x8024967e0
      LISTOP sysseek (KIDS, WANT_LIST, WANT_SCALAR)   0x8024968e8
        OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x802496c00
        UNOP rv2gv (KIDS, SPECIAL, WANT_LIST, WANT_SCALAR)   0x802496f30
          UNOP entersub (KIDS, STACKED, WANT_LIST, WANT_SCALAR)   0x8022d6018
            BINOP null (ex-pp_list) (KIDS, WANT_LIST, WANT_SCALAR)   0x8022d6330
              OpTree pushmark (WANT_LIST, WANT_SCALAR)   0x8022d6660
              UNOP null (ex-pp_rv2cv) (KIDS, WANT_LIST, WANT_SCALAR)  KVSLICE 0x8022d6990
                PADOP gv (WANT_LIST, WANT_SCALAR)   0x8022d6a68
        SVOP const (WANT_LIST, WANT_SCALAR) 10  0x8022d6db0
        SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x8022d6f78
      OpTree padsv (MOD, REF, SPECIAL, WANT_LIST, WANT_SCALAR) $a  0x8022d8078
    COP nextstate (WANT_LIST, WANT_VOID) (eval 183):2  0x8022d8180
    OpTree padsv (MOD, WANT_LIST, WANT_VOID) $fh  0x8022d8288
    COP nextstate (WANT_LIST, WANT_VOID) (eval 183):3  0x8022d8360
    LISTOP sysseek (KIDS, WANT_LIST, WANT_VOID)   0x8022d8438
      OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x8022d8510
      UNOP rv2gv (KIDS, SPECIAL, WANT_LIST, WANT_SCALAR)   0x8022d85b8
        OpTree padsv (WANT_LIST, WANT_SCALAR) $fh  0x8022d8678
      SVOP const (WANT_LIST, WANT_SCALAR) -10  0x8022d8738
      SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x8022d8840
    COP nextstate (WANT_LIST, WANT_VOID) (eval 183):4  0x8022d8918
    LISTOP sysseek (KIDS, WANT_LIST, WANT_SCALAR)   0x8022d8a20
      OpTree null (ex-pp_pushmark) (WANT_LIST, WANT_SCALAR)   0x8022d8af8
      UNOP rv2gv (KIDS, REF, WANT_LIST, WANT_SCALAR)   0x8022d8ba0
        OpTree padsv (WANT_LIST, WANT_SCALAR) $fh  0x8022d8c60
      SVOP const (WANT_LIST, WANT_SCALAR) 0  0x8022d8d20
      SVOP const (SPECIAL, WANT_LIST, WANT_SCALAR) ''  0x8022d8e28
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
Use of uninitialized value $string in pattern match (m//) at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 866.
Use of uninitialized value $string in concatenation (.) or string at /usr/home/jkeenan/.cpanm/work/1645814451.17065/Devel-Chitin-0.18/blib/lib/Devel/Chitin/OpTree.pm line 871.
    ok 24 - t/20-optree-deparse/12-i-o/syswrite
    ok 25 - t/20-optree-deparse/12-i-o/tell
    ok 26 - t/20-optree-deparse/12-i-o/telldir
    ok 27 - t/20-optree-deparse/12-i-o/truncate
    ok 28 - t/20-optree-deparse/12-i-o/warn
    ok 29 - t/20-optree-deparse/12-i-o/write
not ok 12 - Subtest: t/20-optree-deparse/12-i-o

# Failed test 'Subtest: t/20-optree-deparse/12-i-o'
# at t/20-optree-deparse.t line 43.

[snip]

Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/29 subtests 

Test Summary Report
-------------------
t/20-optree-deparse.t (Wstat: 256 Tests: 29 Failed: 1)
  Failed test:  12
  Non-zero exit status: 1
Files=1, Tests=29,  0 wallclock secs ( 0.07 usr  0.02 sys +  0.39 cusr  0.05 csys =  0.52 CPU)
Result: FAIL
jkeenan commented 2 years ago

@nwc10, can you take a look? ^^ Thanks.

Leont commented 2 years ago

The Storable::AMF tests literally do stuff like this:

$a = 1;
$b = "".$a;
$var = "Int++";
ok(    is_amf_string($a)                         , "$var converted  is   a string"  );

Clearly, that is no longer going to work, and it never should have had such an expectation. Looks like these tests should be modified to gracefully handle the new situation (or possibly just be deleted)

Leont commented 2 years ago

Devel::Chitin is using B for some deep introspection. It's calling $sv->PV on B::SV values that no longer have a PV.

I'm not sure how to handle this, possibly this needs a fix in B.

jkeenan commented 2 years ago

The Storable::AMF tests literally do stuff like this:

$a = 1;
$b = "".$a;
$var = "Int++";
ok(    is_amf_string($a)                         , "$var converted  is   a string"  );

Clearly, that is no longer going to work, and it never should have had such an expectation. Looks like these tests should be modified to gracefully handle the new situation (or possibly just be deleted)

I think we're going to need some stock explanation of why and how downstream code must change to adapt to the (1 character!) change in https://github.com/Perl/perl5/commit/bb5bc9. An explanation that an ordinary Perl user -- someone without a deep understanding of XS or sv.c can understand. An explanation that we can paste into a bug ticket for an affected CPAN module.

Any suggestions?

jkeenan commented 2 years ago

Devel::Chitin is using B for some deep introspection. It's calling $sv->PV on B::SV values that no longer have a PV.

I'm not sure how to handle this, possibly this needs a fix in B.

@iabyn, @tonycoz, @atoomic, @toddr -- can you take a look ^^?

toddr commented 2 years ago

Devel::Chitin is using B for some deep introspection. It's calling $sv->PV on B::SV values that no longer have a PV.

I'm not sure how to handle this, possibly this needs a fix in B.

The code in question appears to be https://github.com/brummett/Devel-Chitin/search?q=%24sv-%3Epv

I can't tell if it's in a higher call but it seems like they need to do something like:

$sv->FLAGS & B::SVf_POK or die("This isn't a PV");

Before they can safely do $sv->PV. They're not and so it's failing 🤷🏻

Leont commented 2 years ago

I can't tell if it's in a higher call but it seems like they need to do something like:

 $sv->FLAGS & B::SVf_POK or die("This isn't a PV");

Before they can safely do $sv->PV. They're not and so it's failing 🤷🏻

That is one approach. The other would be to also return a string if SVp_POK is set. I'm currently leaning to that, but I'm maybe I'm not seeing some repercussion of that.

jkeenan commented 2 years ago

Also affected: BRUMMETT/Devel-Chitin-0.18.tar.gz

Ticket opened upstream at RT 141554.

jkeenan commented 2 years ago

Description

Commit v5.35.8-110-gbb5bc97fde/da1c (bb5bc9) stopped also GRIAN/Storable-AMF-1.23.tar.gz from succeeding with the test suite. We had this commit already in #19450 ; that issue has been closed, but the fix provided over there does not help for Storable::AMF 1.23 ; you can already find sample fail reports on http://matrix.cpantesters.org/?dist=Storable-AMF+1.23

Ticket created upstream at RT 141555.

Thank you very much. Jim Keenan

toddr commented 2 years ago

I can't tell if it's in a higher call but it seems like they need to do something like:

 $sv->FLAGS & B::SVf_POK or die("This isn't a PV");

Before they can safely do $sv->PV. They're not and so it's failing 🤷🏻

That is one approach. The other would be to also return a string if SVp_POK is set. I'm currently leaning to that, but I'm maybe I'm not seeing some repercussion of that.

Right. My problem is not what flag so much as it's unclear if it was a mistake to ever enter the sub if it wasn't set and so the problem is in the caller, not that sub.

nwc10 commented 2 years ago

I can't tell if it's in a higher call but it seems like they need to do something like:

 $sv->FLAGS & B::SVf_POK or die("This isn't a PV");

Before they can safely do $sv->PV. They're not and so it's failing 🤷🏻

That is one approach. The other would be to also return a string if SVp_POK is set. I'm currently leaning to that, but I'm maybe I'm not seeing some repercussion of that.

Right. My problem is not what flag so much as it's unclear if it was a mistake to ever enter the sub if it wasn't set and so the problem is in the caller, not that sub.

It's a bug in Devel::Chitin

You can expose it with earlier versions of perl with something like:

use constant foo => do { my $val = "Four"; $val = 4; $val };
my $q = foo;

The code assumes that anything that is SVt_PV or a subclass must have a string. it can't cope with a PVIV that is actually an IV. That code above looks like this:

$ perl -MO=Concise -e 'use constant foo => do { my $val = "Four"; $val = 4; $val }; my $q = foo'
6  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter v ->2
2     <;> nextstate(main 190 -e:1) v:{ ->3
5     <2> sassign vKS/2 ->6
3        <$> const[PVIV 4] s*/FOLD ->4
4        <0> padsv[$q:190,191] sRM*/LVINTRO ->5
-e syntax OK

ie const[PVIV 4] not the [IV 4] you'd see from something sane:

$ perl -MO=Concise -e 'use constant foo => 4; my $q = foo'
6  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter v ->2
2     <;> nextstate(main 187 -e:1) v:{ ->3
5     <2> sassign vKS/2 ->6
3        <$> const[IV 4] s*/FOLD ->4
4        <0> padsv[$q:187,188] sRM*/LVINTRO ->5
-e syntax OK

The fix (for the above bug, and the regression we see here), is something like this:

$ diff -u lib/Devel/Chitin/OpTree/SVOP.pm~  lib/Devel/Chitin/OpTree/SVOP.pm
--- lib/Devel/Chitin/OpTree/SVOP.pm~    2018-10-25 22:55:02.000000000 +0000
+++ lib/Devel/Chitin/OpTree/SVOP.pm     2022-02-26 07:03:05.549223734 +0000
@@ -6,6 +6,8 @@
 use strict;
 use warnings;

+use B qw(SVf_IOK SVf_NOK SVf_POK);
+
 sub pp_const {
     my $self = shift;
     my %params = @_;
@@ -19,13 +21,12 @@
         for (my $mg = $sv->MAGIC; $mg; $mg = $mg->MOREMAGIC) {
             return $mg->PTR if $mg->TYPE eq 'V';
         }
-
-    } elsif ($sv->isa('B::PV')) {
+    } elsif ($sv->FLAGS & SVf_POK) {
         return $self->_quote_sv($sv, %params);
-    } elsif ($sv->isa('B::NV')) {
-        return $sv->NV;
-    } elsif ($sv->isa('B::IV')) {
+    } elsif ($sv->FLAGS & SVf_IOK) {
         return $sv->int_value;
+    } elsif ($sv->FLAGS & SVf_NOK) {
+        return $sv->NV;
     } elsif ($sv->isa('B::SPECIAL')) {
         '<???pp_const B::SPECIAL ' .  $B::specialsv_name[$$sv] . '>';

(adapted from something in B, which clearly is more robust)

I was working on that last night, but it got too late and I hadn't quite nailed it.

The cause of the confusion is that effectively the regression tests do this:

use Fcntl qw(SEEK_SET SEEK_CUR SEEK_END); # omit

BEGIN {
    my %seek_flags = (
        SEEK_SET() => 'SEEK_SET',
        SEEK_CUR() => 'SEEK_CUR',
        SEEK_END() => 'SEEK_END',
    );
}

my $a = seek(F, 10, SEEK_CUR);
my $fh;
seek($fh, -10, SEEK_END);
seek(*$fh, 0, SEEK_SET)

and constructing that hash (in lib/Devel/Chitin/OpTree/LISTOP.pm) causes the constants in Fcntl to be converted from IV to PVIV as a side effect of "reading" the numbers as strings.

Which we've just changed so as not to set SVf_POK

I might not get time to look at this further for at least 4 days.

demerphq commented 2 years ago

On Sat, 26 Feb 2022 at 00:23, Leon Timmermans @.***> wrote:

I can't tell if it's in a higher call but it seems like they need to do something like:

$sv->FLAGS & B::SVf_POK or die("This isn't a PV");

Before they can safely do $sv->PV. They're not and so it's failing 🤷🏻

That is one approach. The other would be to also return a string if POKp is set. I'm currently leaning to that, but I'm maybe I'm not seeing some repercussion of that.

That makes a lot more sense to me.

Yves

-- perl -Mre=debug -e "/just|another|perl|hacker/"

nwc10 commented 2 years ago

The Storable::AMF tests literally do stuff like this:

$a = 1;
$b = "".$a;
$var = "Int++";
ok(    is_amf_string($a)                         , "$var converted  is   a string"  );

Clearly, that is no longer going to work, and it never should have had such an expectation. Looks like these tests should be modified to gracefully handle the new situation (or possibly just be deleted)

If you look at Grian/Storable-AMF@70ff9715265e639a034c44b01201ab7c5eda0041 you'll see how it handled the previous analogous change, when stringification of NVs stopped setting SVf_POK:

@@ -55,7 +55,8 @@ $a = 1.0;
 # DoubleVar
 $a = 1.0;
 $b = "$a";
 $var = "Double";

-ok(   is_amf_string($a)                         , "$var converted  is   a string"  );
+my %string_number=(0=>1,2=>1);
+ok(    $string_number{ byte_amf_string($a)} , "$var converted  is   a string"  );
 ok( !  is_amf_string(0+$a)              , "$var 0+converted     is double" );
 ok( !  is_amf_string(0.0+$a,$nop), "$var 0.0+converted   is double" );
 ok( !  is_amf_string($a, $nop)   , "$var converted again is double" );

with:

sub is_amf_string{
        ord( freeze( $_[0], $_[1]||0 )) == 2;
}
sub byte_amf_string{
        ord( freeze( $_[0], $_[1]||0 ));

That change is clearly not the world's most robust piece of code logic.

I think I get the intent of the tests, but (again) I have run out of time for now. I think that the intent is if you have this:

$ perl -MDevel::Peek -e '$foo = "42"; sqrt $foo; Dump $foo'
SV = PVNV(0x55b3fc452240) at 0x55b3fc4801a0
  REFCNT = 1
  FLAGS = (NOK,POK,IsCOW,pNOK,pPOK)
  IV = 0
  NV = 42
  PV = 0x55b3fc4c18d0 "42"\0
  CUR = 2
  LEN = 10
  COW_REFCNT = 1

1) does it stringify as a number, or a string? (default options) 2) and does setting the option change that

and analogous tests for PVIVs which have both POK and IOK (the ones failing today)

There were then later tests added for "if I used the value in arithmetic does it change?" which I think are superfluous to what the module tests.

So I think that the correct fix is to reduce the module's regression tests down to what it actually needs to test the module (not perl's behaviours, that have changed). ie start with a string, and do an integer or a floating point operation on it to force IOK or NOK on as well, but with POK still set.

But I've run out of time (for the next few days at least) to figure this out properly.

jkeenan commented 2 years ago

Also affected: BRUMMETT/Devel-Chitin-0.18.tar.gz

Ticket opened upstream at RT 141554.

The upstream maintainer has addressed this issue and has released a new version of Devel-Chitin to CPAN. So this ticket is now solely concerned with Storable-AMF. The upstream ticket for that is RT 141555.

rjbs commented 2 years ago

I don't believe this should block the release of v5.36.0 but have not (yet) added a non-5.36-blocker label, as I'm hoping we'll see a new release.

@nwc10 What do you think about the value of adding a "how this may break your slightly wrong code" section to perldelta?

nwc10 commented 2 years ago

What do you think about the value of adding a "how this may break your slightly wrong code" section to perldelta?

I'm really not sure if there is a useful wording. Since the merge we've uncovered (I think) 2 things that were bugs in the core that needed fixing. The rest grouped as

  1. your code was explicitly using B to read flags (and the module in question was already subtly buggy - we just exposed it directly in an existing test)
  2. your code is sensitive to the length of the serealised form of a data structure (and in this case it exposed a bug in that code, or maybe the tests)
  3. your code is trying to test XS code which handles strings and numbers differently, and your approach to creating values that meet your tests expectations no longer exactly matches how the core does things

but >50% of these were "we exposed existing bugs in your code"

and the real problem is that it's "your code, or any of your dependency chain". I'm not sure how to offer specific advice that's tight enough to be useful, but simulateously suggests that all wierd upgrade woes are caused by this (because most won't be)

I guess it's "your or your dependencies used B directly to read flags, or you or your dependencies rely exactly on how your serealiser stores data structures" but as much "and they were fragile and you didn't know this before"

rjbs commented 2 years ago

I am now marking this as non-blocking.

jkeenan commented 2 years ago

I believe we should close this ticket, for the following reasons:

  1. Storable-AMF is the only upstream distro still needing action.
  2. The upstream maintainer has not responded to the bug ticket I filed in February about the problem.
  3. There has not been a new release of this distribution in more than 6 years.

Hence, there's nothing more that P5P can do at this time. I'm self-assigning this ticket for the purpose of closing it within 14 days unless someone wants to carry the discussion forward.

jkeenan commented 2 years ago

I believe we should close this ticket, for the following reasons:

1. Storable-AMF is the only upstream distro still needing action.

2. The upstream maintainer has not responded to the [bug ticket](https://rt.cpan.org/Ticket/Display.html?id=141555) I filed in February about the problem.

3. There has not been a new release of this distribution in more than 6 years.

Hence, there's nothing more that P5P can do at this time. I'm self-assigning this ticket for the purpose of closing it within 14 days unless someone wants to carry the discussion forward.

No objection heard to closing this ticket; doing so now.