code review 5528053: spec: pointer comparison for pointers to 0-sized variables

spec: pointer comparison for pointers to 0-sized variables

- define "0-sized"
- add clarifying sentence to pointer comparison
- removed notion "location" which was used only in pointer comparisons
  and which was never defined

Fixes issue 2620.

[gri, 2012-01-09 22:13:05]

Hello golang-dev@googlegroups.com,

I'd like you to review this change to
https://go.googlecode.com/hg/

[r, 2012-01-09 22:24:37]

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html
File doc/go_spec.html (right):

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html#newcode2960
doc/go_spec.html:2960: Pointers to distinct <a
href="#Size_and_alignment_guarantees">0-sized</a> variables may or may not be
equal.
s/0-sized/zero-size/g

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html#newcode5352
doc/go_spec.html:5352: A struct or array type has size 0 if it contains no
fields (or elements, respectively) that have a size > 0. Two distinct 0-sized
variables may have the same address in memory.
&gt;

or better to write it out: have a size greater than zero.

[gri, 2012-01-09 22:29:58]

PTAL

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html
File doc/go_spec.html (right):

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html#newcode2960
doc/go_spec.html:2960: Pointers to distinct <a
href="#Size_and_alignment_guarantees">0-sized</a> variables may or may not be
equal.
On 2012/01/09 22:24:37, r wrote:
> s/0-sized/zero-size/g

Done.

http://codereview.appspot.com/5528053/diff/5002/doc/go_spec.html#newcode5352
doc/go_spec.html:5352: A struct or array type has size 0 if it contains no
fields (or elements, respectively) that have a size > 0. Two distinct 0-sized
variables may have the same address in memory.
On 2012/01/09 22:24:37, r wrote:
> &gt;
> 
> or better to write it out: have a size greater than zero.
> 

Done.

[r, 2012-01-09 22:34:49]

LGTM but wait for others.

[rsc, 2012-01-09 22:43:00]

LGTM

If "variable" is the right name, then lgtm.
It said "location" before because if you have

var x struct { y, z int }

&x.y and &x.z are different pointers even though
they both point into the variable x.

[gri, 2012-01-09 22:47:52]

On Mon, Jan 9, 2012 at 2:43 PM,  <rsc@golang.org> wrote:
> LGTM
>
> If "variable" is the right name, then lgtm.
> It said "location" before because if you have
>
> var x struct { y, z int }
>
> &x.y and &x.z are different pointers even though
> they both point into the variable x.

I think this our long-standing issue we have with the notion of what a
variable is. In my mind it's a memory location that can hold different
values of a given type. Some variables are named explicitly (x), some
are named with an index (a[i]), and some are anonymous and accessed
via a pointer indirection (*p). Others prefer to call this notion an
L-value. So far we have always avoided both notions by doing a
delicate dance around it. Maybe it's time to bite the bullet (in a
different CL) and define a term. It might simplify some other parts of
the spec.

- gri

>
>
> http://codereview.appspot.com/5528053/

[iant, 2012-01-10 00:35:37]

LGTM

[gri, 2012-01-10 00:54:27]

*** Submitted as http://code.google.com/p/go/source/detail?r=9fac7377fc7e ***

spec: pointer comparison for pointers to 0-sized variables

- define "0-sized"
- add clarifying sentence to pointer comparison
- removed notion "location" which was used only in pointer comparisons
  and which was never defined

Fixes issue 2620.

R=r, rsc, iant
CC=golang-dev
http://codereview.appspot.com/5528053