code review 93550043: math/big: implement Rat.Float32

math/big: implement Rat.Float32

Pending CL 101750048.
For submission after the 1.3 release.

Fixes issue 8065.

[gri, 2014-05-22 22:58:44]

Hello adonovan@google.com (cc: golang-codereviews@googlegroups.com),

I'd like you to review this change to
https://code.google.com/p/go/

[adonovan, 2014-05-23 02:23:40]

LG, but needs tests.

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go
File src/pkg/math/big/rat.go (right):

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
src/pkg/math/big/rat.go:67: // quotToFloat32 returns the non-negative IEEE 754
double-precision
"single-precision"

but you could replace four words by "float32".  Ditto for 64.

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
src/pkg/math/big/rat.go:142: exp = 2 - Ebias // -127 + 2 TODO(gri) why?
It's equivalent to exp += shift.  Here's the derivation.

exp += shift
exp = exp + shift
exp = exp + Emin - (exp - 1)
exp = Emin + 1
exp = (1 - Ebias) + 1
exp = 2 - Ebias

[gri, 2014-05-28 20:10:42]

PTAL

Also added tests.

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go
File src/pkg/math/big/rat.go (right):

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
src/pkg/math/big/rat.go:67: // quotToFloat32 returns the non-negative IEEE 754
double-precision
On 2014/05/23 02:23:40, adonovan wrote:
> "single-precision"
> 
> but you could replace four words by "float32".  Ditto for 64.

Done.

https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
src/pkg/math/big/rat.go:142: exp = 2 - Ebias // -127 + 2 TODO(gri) why?
On 2014/05/23 02:23:40, adonovan wrote:
> It's equivalent to exp += shift.  Here's the derivation.
> 
> exp += shift
> exp = exp + shift
> exp = exp + Emin - (exp - 1)
> exp = Emin + 1
> exp = (1 - Ebias) + 1
> exp = 2 - Ebias 

Done.

[adonovan, 2014-05-28 20:29:21]

On 2014/05/28 20:10:42, gri wrote:
> PTAL
> 
> Also added tests.
> 
> https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go
> File src/pkg/math/big/rat.go (right):
> 
>
https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
> src/pkg/math/big/rat.go:67: // quotToFloat32 returns the non-negative IEEE 754
> double-precision
> On 2014/05/23 02:23:40, adonovan wrote:
> > "single-precision"
> > 
> > but you could replace four words by "float32".  Ditto for 64.
> 
> Done.
> 
>
https://codereview.appspot.com/93550043/diff/20002/src/pkg/math/big/rat.go#ne...
> src/pkg/math/big/rat.go:142: exp = 2 - Ebias // -127 + 2 TODO(gri) why?
> On 2014/05/23 02:23:40, adonovan wrote:
> > It's equivalent to exp += shift.  Here's the derivation.
> > 
> > exp += shift
> > exp = exp + shift
> > exp = exp + Emin - (exp - 1)
> > exp = Emin + 1
> > exp = (1 - Ebias) + 1
> > exp = 2 - Ebias 
> 
> Done.

LGTM

[gri, 2014-06-11 16:10:52]

*** Submitted as https://code.google.com/p/go/source/detail?r=cb1af30a996c ***

math/big: implement Rat.Float32

Pending CL 101750048.
For submission after the 1.3 release.

Fixes issue 8065.

LGTM=adonovan
R=adonovan
CC=golang-codereviews
https://codereview.appspot.com/93550043

[gobot, 2014-06-11 16:57:47]

This CL appears to have broken the darwin-386 builder.
See http://build.golang.org/log/a6d89a79deaa5fe8001564491cbc95d862097fbb