large right-shift of negative loses sign

[p6rt]

Migrated from rt.perl.org#126942 (status was 'resolved')

Searchable as RT126942$

[p6rt]

From zefram@fysh.org

Long right shifts (longer than operand length) of negative integers
should always produce -1, but in Rakudo after some length they stop
doing that and switch to an incorrect output​:

-123 +> 8
-1
-123 +> 31
-1
-123 +> 32
0
-123 +> (1 +< 60)
0

Some very long right shifts then run into [perl #​126940] and [perl
#​126941], but those are separate issues.

-zefram

[p6rt]

From @lizmat

On 17 Dec 2015, at 03​:41, Zefram (via RT) <perl6-bugs-followup@​perl.org> wrote​:

# New Ticket Created by Zefram
# Please include the string​: [perl #​126942]
# in the subject line of all future correspondence about this issue.
# <URL​: https://rt-archive.perl.org/perl6/Ticket/Display.html?id=126942 >

Long right shifts (longer than operand length) of negative integers
should always produce -1, but in Rakudo after some length they stop
doing that and switch to an incorrect output​:

-123 +> 8
-1
-123 +> 31
-1
-123 +> 32
0
-123 +> (1 +< 60)
0

Some very long right shifts then run into [perl #​126940] and [perl
#​126941], but those are separate issues.

$ 6 'say -123 +> 32'
-1

Temporarily fixed with 0e24208abfcc67b22a7f , test added with 83057c9b8f97ed9b761f1 , should probably get a better fix at nqp level post Christmas. Or not.

Liz

[p6rt]

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

[p6rt]

From 1parrota@gmail.com

Does a shift value longer than the word length make any sense anyway?

On 12/17/15, Elizabeth Mattijsen <liz@​dijkmat.nl> wrote​:

On 17 Dec 2015, at 03​:41, Zefram (via RT) <perl6-bugs-followup@​perl.org>
wrote​:

# New Ticket Created by Zefram
# Please include the string​: [perl #​126942]
# in the subject line of all future correspondence about this issue.
# <URL​: https://rt-archive.perl.org/perl6/Ticket/Display.html?id=126942 >

Long right shifts (longer than operand length) of negative integers
should always produce -1, but in Rakudo after some length they stop
doing that and switch to an incorrect output​:

-123 +> 8
-1
-123 +> 31
-1
-123 +> 32
0
-123 +> (1 +< 60)
0

Some very long right shifts then run into [perl #​126940] and [perl
#​126941], but those are separate issues.

$ 6 'say -123 +> 32'
-1

Temporarily fixed with 0e24208abfcc67b22a7f , test added with
83057c9b8f97ed9b761f1 , should probably get a better fix at nqp level post
Christmas. Or not.

Liz

[p6rt]

From zefram@fysh.org

Parrot Raiser via RT wrote​:

Does a shift value longer than the word length make any sense anyway?

With bignums, yes it does.

-123 +< 200
-197653379443855803891661337357963000110230968235283518742069248

(Also, 32 isn't really my word length.)

-zefram

[p6rt]

From @zoffixznet

Thanks for the report. This is now fixed.

Fix​:
  MoarVM/MoarVM@362277b79f
  Raku/nqp@9763728215
  rakudo/rakudo@66e8e72c02
  rakudo/rakudo@ef29bb9f41

Tests​:
  Raku/roast@02d1f9d0f6
  rakudo/rakudo@6409ee582e

[p6rt]

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