From eigil@skjaeve.land

my Int @​a = 1, 2;
my int @​b = 1, 2;
say so @​a[-1 ^ 0];
True
say so @​b[-1 ^ 0];
False
say @​a[-1 ^ 0];
Index out of range. Is​: -1, should be in 0..^Inf
  in block <unit> at <unknown file> line 1

say @​b[-1 ^ 0];
one(2, 1)

It seems the native int array treats negative indices in a junction as
relative to the end of the array.

This is Rakudo version 2017.01-58-gd06d7c16b built on MoarVM version 2017.01-13-g4aab5064
implementing Perl 6.c.

From @lizmat

On 11 Apr 2017, at 14​:44, Eigil Skjæveland (via RT) <perl6-bugs-followup@​perl.org> wrote​:

# New Ticket Created by Eigil Skjæveland
# Please include the string​: [perl #​131134]
# in the subject line of all future correspondence about this issue.
# <URL​: https://rt-archive.perl.org/perl6/Ticket/Display.html?id=131134 >

my Int @​a = 1, 2;
my int @​b = 1, 2;
say so @​a[-1 ^ 0];
True
say so @​b[-1 ^ 0];
False
say @​a[-1 ^ 0];
Index out of range. Is​: -1, should be in 0..^Inf
in block <unit> at <unknown file> line 1

say @​b[-1 ^ 0];
one(2, 1)

It seems the native int array treats negative indices in a junction as
relative to the end of the array.

Any expression resulting in a negative value at runtime, exposes this behaviour​:

$ 6 'my int @​a = ^10; my $a = 0; dd @​a[$a-1]’
9

At the moment, this happens because the AT-POS candidate

The RT System itself - Status changed from 'new' to 'open'

From @lizmat

On 11 Apr 2017, at 14​:44, Eigil Skjæveland (via RT) <perl6-bugs-followup@​perl.org> wrote​:

# New Ticket Created by Eigil Skjæveland
# Please include the string​: [perl #​131134]
# in the subject line of all future correspondence about this issue.
# <URL​: https://rt-archive.perl.org/perl6/Ticket/Display.html?id=131134 >

my Int @​a = 1, 2;
my int @​b = 1, 2;
say so @​a[-1 ^ 0];
True
say so @​b[-1 ^ 0];
False
say @​a[-1 ^ 0];
Index out of range. Is​: -1, should be in 0..^Inf
in block <unit> at <unknown file> line 1

say @​b[-1 ^ 0];
one(2, 1)

It seems the native int array treats negative indices in a junction as
relative to the end of the array.

Any expression resulting in a negative value at runtime, exposes this behaviour​:

$ 6 'my int @​a = ^10; my $a = 0; dd @​a[$a-1]’
9

At the moment, this happens because the AT-POS candidate directly uses nqp​::atposref_i for performance reasons.

Exactly for that reason, I don’t think we should fix this at the Perl 6 level, because we would lose a *lot* of performance if we did.

So, *if* we consider this to be a bug (and personally, I think it is), the index checking should happen at the VM level, rather than at the Perl 6 level.

From @lizmat

On 14 Apr 2017, at 10​:29, Elizabeth Mattijsen <liz@​dijkmat.nl> wrote​:

On 11 Apr 2017, at 14​:44, Eigil Skjæveland (via RT) <perl6-bugs-followup@​perl.org> wrote​:

# New Ticket Created by Eigil Skjæveland
# Please include the string​: [perl #​131134]
# in the subject line of all future correspondence about this issue.
# <URL​: https://rt-archive.perl.org/perl6/Ticket/Display.html?id=131134 >

my Int @​a = 1, 2;
my int @​b = 1, 2;
say so @​a[-1 ^ 0];
True
say so @​b[-1 ^ 0];
False
say @​a[-1 ^ 0];
Index out of range. Is​: -1, should be in 0..^Inf
in block <unit> at <unknown file> line 1

say @​b[-1 ^ 0];
one(2, 1)

It seems the native int array treats negative indices in a junction as
relative to the end of the array.

Any expression resulting in a negative value at runtime, exposes this behaviour​:

$ 6 'my int @​a = ^10; my $a = 0; dd @​a[$a-1]’
9

At the moment, this happens because the AT-POS candidate directly uses nqp​::atposref_i for performance reasons.

Exactly for that reason, I don’t think we should fix this at the Perl 6 level, because we would lose a *lot* of performance if we did.

Hmmm… the loss in performance appears to be in the 10% range, so maybe not as bad as I anticipated.

Liz