From @AlexDaniel

Two tests in Data​::MessagePack started to fail​:

Test Summary Report

t/202-unpack-int.t (Wstat​: 256 Tests​: 40 Failed​: 1)
  Failed test​: 35
  Non-zero exit status​: 1
t/302-stream-unpack-int.t (Wstat​: 256 Tests​: 23 Failed​: 1)
  Failed test​: 21
  Non-zero exit status​: 1

MessagePack​::Class depends on Data​::MessagePack and therefore is not installable also.

Committable output for 202-unpack-int.t on 2017.08 and HEAD​: https://gist.github.com/50eab41e808e812c7d555f45272dc2a5

Bisectable points at rakudo/rakudo@9b42484

Module repo​: https://github.com/pierre-vigier/Perl6-Data-MessagePack

It could very well be that it is not a regression, but that the module was relying on buggy behavior. I think having a ticket is reasonable until we get a better answer. Any info is welcome.

From @AlexDaniel

This is definitely an issue in MoarVM, and it existed since 2014.02 but for some reason it was unnoticeable before recent changes.

Code​:
my $value = -0x7FFFFFFF - 1;
say $value;
say -2147483648;
say $value == -2147483648;

Result (MoarVM)​:
-2147483648
-2147483648
False

Result (JVM)​:
-2147483648
-2147483648
True

Maybe this can help​:
Code​:
my $value = -0x7FFFFFFF - 1;
say $value;
say $value == -18446744071562067968

Result (MoarVM)​:
-2147483648
True

On 2017-09-13 19​:49​:24, alex.jakimenko@​gmail.com wrote​:

Two tests in Data​::MessagePack started to fail​:

Test Summary Report
-------------------
t/202-unpack-int.t (Wstat​: 256 Tests​: 40 Failed​: 1)
Failed test​: 35
Non-zero exit status​: 1
t/302-stream-unpack-int.t (Wstat​: 256 Tests​: 23 Failed​: 1)
Failed test​: 21
Non-zero exit status​: 1

MessagePack​::Class depends on Data​::MessagePack and therefore is not
installable also.

Committable output for 202-unpack-int.t on 2017.08 and HEAD​:
https://gist.github.com/50eab41e808e812c7d555f45272dc2a5

Bisectable points at
rakudo/rakudo@9b42484

Module repo​: https://github.com/pierre-vigier/Perl6-Data-MessagePack

It could very well be that it is not a regression, but that the module
was relying on buggy behavior. I think having a ticket is reasonable
until we get a better answer. Any info is welcome.

From @AlexDaniel

The issue seems to be resolved in MoarVM/MoarVM@9c5fed7

Tests needed.

When adding tests for this, please cover this case also​:
say (-0x7FFFFFFF - 1).base(16) # should probably be -80000000 but it erroneously gives -FFFFFFFF80000000

On 2017-09-16 10​:46​:51, alex.jakimenko@​gmail.com wrote​:

This is definitely an issue in MoarVM, and it existed since 2014.02
but for
some reason it was unnoticeable before recent changes.

Code​:
my $value = -0x7FFFFFFF - 1;
say $value;
say -2147483648;
say $value == -2147483648;

Result (MoarVM)​:
-2147483648
-2147483648
False

Result (JVM)​:
-2147483648
-2147483648
True

Maybe this can help​:
Code​:
my $value = -0x7FFFFFFF - 1;
say $value;
say $value == -18446744071562067968

Result (MoarVM)​:
-2147483648
True

On 2017-09-13 19​:49​:24, alex.jakimenko@​gmail.com wrote​:

Two tests in Data​::MessagePack started to fail​:

Test Summary Report
-------------------
t/202-unpack-int.t (Wstat​: 256 Tests​: 40 Failed​: 1)
Failed test​: 35
Non-zero exit status​: 1
t/302-stream-unpack-int.t (Wstat​: 256 Tests​: 23 Failed​: 1)
Failed test​: 21
Non-zero exit status​: 1

MessagePack​::Class depends on Data​::MessagePack and therefore is not
installable also.

Committable output for 202-unpack-int.t on 2017.08 and HEAD​:
https://gist.github.com/50eab41e808e812c7d555f45272dc2a5

Bisectable points at

rakudo/rakudo@9b42484

Module repo​: https://github.com/pierre-vigier/Perl6-Data-MessagePack

It could very well be that it is not a regression, but that the
module
was relying on buggy behavior. I think having a ticket is reasonable
until we get a better answer. Any info is welcome.

From @zoffixznet

Tests in Raku/roast#345

@zoffixznet - Status changed from 'new' to 'resolved'