Restore exact old semantics of `decodeFloat`
ClosedPublic
Actions

Authored by hvr on Nov 18 2014, 1:39 AM.

Details

Reviewers

luite
austin
ekmett
rwbarton

Commits

rGHCDIFFe2af452cd533: Restore exact old semantics of `decodeFloat`
rGHCe2af452cd533: Restore exact old semantics of `decodeFloat`

Trac Issues

#9810

Summary

integer-gmp2 uses a new 64bit-based primop
introduced in b62bd5ecf3be421778e4835010b6b334e95c5a56.

However, the returned values differ for exceptional IEEE values:

Previous (expected) semantics:

> decodeFloat (-1/0)
(-4503599627370496,972)

> decodeFloat (1/0)
(4503599627370496,972)

> decodeFloat (0/0)
(-6755399441055744,972)

Currently (broken) semantics:

> decodeFloat (-1/0 :: Double)
(-9223372036854775808,-53)

> decodeFloat (1/0 :: Double)
(-9223372036854775808,-53)

> decodeFloat (0/0 :: Double)
(-9223372036854775808,-53)

This patch reverts to the old expected semantics.

We'll revisit the implementation during GHC 7.11 development.

This should address Trac #9810

Diff Detail

Repository

rGHC Glasgow Haskell Compiler

Lint

Automatic diff as part of commit; lint not applicable.

Unit

Automatic diff as part of commit; unit tests not applicable.

hvr updated this revision to Diff 1483.Nov 18 2014, 1:39 AM

hvr retitled this revision from to Restore exact old semantics of `decodeFloat`.

hvr updated this object.

hvr edited the test plan for this revision. (Show Details)

hvr added reviewers: luite, rwbarton, ekmett.

hvr updated the Trac tickets for this revision.

Herald added a reviewer: austin. · View Herald TranscriptNov 18 2014, 1:39 AM

Herald added subscribers: carter, thomie. · View Herald Transcript

Harbormaster failed remote builds in B1966: Diff 1483!Nov 18 2014, 2:05 AM

Here's some interesting semantics of NaN:

> let dumpIEEE x = putStrLn $ "0x" ++ showHex (unsafeCoerce# (x::Double) :: Word64) ""

> dumpIEEE (1/0)
0x7ff0000000000000
> dumpIEEE (-1/0)
0xfff0000000000000
> dumpIEEE (0/0)
0xfff8000000000000
> dumpIEEE (-0/0)
0x7ff8000000000000

> decodeFloat (0/0 :: Double)
(-6755399441055744,972)

> decodeFloat (-0/0 :: Double)
(6755399441055744,972)

Note how 0/0 has a negative sign (while -0/0 is a positive signed NaN) above, however:

> signum (0/0 :: Double)
-1.0
> signum (-0/0 :: Double)
-1.0

This makes we wonder whether

0/0 shouldn't rather be positive (and -0/0 be negative), and
the following shall hold: signum (0/0) == 1 and signum (-0/0) == -1, as well as
signum (abs x) == 1 and signum (negate x) == negate (signum x) for x in {0/0, -0/0}.

In D486#12484, @hvr wrote:
Note how 0/0 has a negative sign (while -0/0 is a positive signed NaN) above, however:
> signum (0/0 :: Double)
 -1.0
> signum (-0/0 :: Double)
 -1.0
This makes we wonder whether

0/0 shouldn't rather be positive (and -0/0 be negative), and

the following shall hold: signum (0/0) == 1 and signum (-0/0) == -1, as well as

signum (abs x) == 1 and signum (negate x) == negate (signum x) for x in {0/0, -0/0}.

Were you testing with HEAD or 7.8? I thought we fixed this in Trac #7858 (defining signum of NaN to be NaN).

In D486#12563, @rwbarton wrote:

Were you testing with HEAD or 7.8? I thought we fixed this in Trac #7858 (defining signum of NaN to be NaN).

oh... ok, (fixed) GHC HEAD does this btw (it does indeed pass-thru NaN, including its sign):

> decodeFloat (signum (-0/0))
(6755399441055744,972)

> decodeFloat (signum (0/0))
(-6755399441055744,972)

LGTM.

This revision is now accepted and ready to land.Nov 18 2014, 12:03 PM

LGTM

hrm, how come on my 7.8.3 ghci i'm seeing the the following

Prelude> decodeFloat (signum (-0/0 :: Double))
(-4503599627370496,-52)
Prelude> decodeFloat (signum (0/0 :: Double))
(-4503599627370496,-52)

yep

Let's also add a test that fromRational (toRational x) == x for x <- [1/0, -1/0], since GHC depends on it.

In D486#12701, @carter wrote:
hrm, how come on my 7.8.3 ghci i'm seeing the the following
Prelude> decodeFloat (signum (-0/0 :: Double))
(-4503599627370496,-52)
Prelude> decodeFloat (signum (0/0 :: Double))
(-4503599627370496,-52)

because you're effectively evaluating decodeFloat (-1.0 :: Double)

rebase & testcase

Harbormaster completed remote builds in B2035: Diff 1544.Nov 19 2014, 5:08 AM

Closed by commit rGHCe2af452cd533: Restore exact old semantics of `decodeFloat` (authored by hvr). · Explain WhyNov 19 2014, 5:14 AM

This revision was automatically updated to reflect the committed changes.

Revision Contents
Changeset List

		Path
M		rts/StgPrimFloat.c (13 lines)
M		testsuite/.gitignore (1 line)
A	M	testsuite/tests/numeric/should_run/T9810.hs (25 lines)
A	M	testsuite/tests/numeric/should_run/T9810.stdout (14 lines)
M		testsuite/tests/numeric/should_run/all.T (1 line)

Diff	ID	Base	Description	Created	Lint	Unit
Base			Base
Diff 1	1483	b9096df		Nov 18 2014, 1:39 AM	★	★
Diff 2	1544	4224466	rebase & testcase	Nov 19 2014, 4:38 AM	★	★
Diff 3	1546	4224466	Commit rGHCe2af452cd533778c5447719c59429d72bb1fe00d	Nov 19 2014, 4:37 AM	★	★