code review 1699045: reflect: add Type.Bits method, add tags to prohibit con...

reflect: add Type.Bits method, add tags to prohibit conversions
gob: substitute slice for map

[rsc, 2010-06-21 17:40:53]

Hello r (cc: golang-dev@googlegroups.com),

I'd like you to review this change.

[r2, 2010-06-21 18:02:13]

http://codereview.appspot.com/1699045/diff/7001/8007
File src/pkg/reflect/type.go (right):

http://codereview.appspot.com/1699045/diff/7001/8007#newcode40
src/pkg/reflect/type.go:40: // will all be different, so that the client cannot
convert
too may "so that"s.

http://codereview.appspot.com/1699045/diff/7001/8007#newcode41
src/pkg/reflect/type.go:41: // from, say, *UintType to *FloatType.
this is a nice trick. does the spec say anything about it?

http://codereview.appspot.com/1699045/diff/7001/8007#newcode226
src/pkg/reflect/type.go:226: Bits() int
i debate with myself whether it should be uint, probably not, even though it
will almost always be used as a shift

[rsc, 2010-06-21 19:37:36]

> http://codereview.appspot.com/1699045/diff/7001/8007#newcode40
> src/pkg/reflect/type.go:40: // will all be different, so that the client
> cannot convert
> too may "so that"s.

reworded.

> http://codereview.appspot.com/1699045/diff/7001/8007#newcode41
> src/pkg/reflect/type.go:41: // from, say, *UintType to *FloatType.
> this is a nice trick. does the spec say anything about it?

not yet; in another cl.

> http://codereview.appspot.com/1699045/diff/7001/8007#newcode226
> src/pkg/reflect/type.go:226: Bits() int
> i debate with myself whether it should be uint, probably not, even
> though it will almost always be used as a shift

right.  if i did change it i'd have to change strconv.AtofN and strconv.FtoaN.

[r2, 2010-06-21 19:39:05]

LGTM

[rsc, 2010-06-21 20:19:34]

*** Submitted as http://code.google.com/p/go/source/detail?r=47185e9a589b ***

reflect: add Type.Bits method, add tags to prohibit conversions
gob: substitute slice for map

R=r
CC=golang-dev
http://codereview.appspot.com/1699045