code review 1687047: spec: len(x) only constant if x is constant

spec: restrict when len(x) is constant

[rsc, 2010-07-01 21:42:10]

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

I'd like you to review this change.

[rsc1, 2010-07-01 21:44:47]

-yuri, +ken

[rsc, 2010-07-01 23:17:32]

Hello gri, iant, ken2, r (cc: golang-dev@googlegroups.com),

Please take another look.

[r2, 2010-07-01 23:23:09]

seems good. gri would say "denoting" for "naming" in the new paragraph.

-rob

[rsc, 2010-07-01 23:26:10]

done

On Thu, Jul 1, 2010 at 16:23, Rob 'Commander' Pike <r@google.com> wrote:
> seems good. gri would say "denoting" for "naming" in the new paragraph.
>
> -rob
>
>

[gri, 2010-07-01 23:34:06]

fyi

http://codereview.appspot.com/1687047/diff/11001/12001
File doc/go_spec.html (right):

http://codereview.appspot.com/1687047/diff/11001/12001#newcode533
doc/go_spec.html:533: <code>cap</code> or <code>len</code> applied to a some
expressions of array type,
s/to a some/to/

http://codereview.appspot.com/1687047/diff/11001/12001#newcode533
doc/go_spec.html:533: <code>cap</code> or <code>len</code> applied to a some
expressions of array type,
does this need a reference to the section on len, cap?

[rsc, 2010-07-01 23:37:31]

ptal

> http://codereview.appspot.com/1687047/diff/11001/12001#newcode533
> doc/go_spec.html:533: <code>cap</code> or <code>len</code> applied to a
> some expressions of array type,
> s/to a some/to/

changed to "to some".  some is important; it's not all

> http://codereview.appspot.com/1687047/diff/11001/12001#newcode533
> doc/go_spec.html:533: <code>cap</code> or <code>len</code> applied to a
> some expressions of array type,
> does this need a reference to the section on len, cap?

added reference; see what you think.

[iant, 2010-07-02 00:04:46]

LGTM

[gri, 2010-07-02 00:14:49]

http://codereview.appspot.com/1687047/diff/17001/18001
File doc/go_spec.html (right):

http://codereview.appspot.com/1687047/diff/17001/18001#newcode4392
doc/go_spec.html:4392: denoting an array.
I thought we also allowed pointers?

[rsc, 2010-07-02 00:15:48]

whoops; fixed.

On Thu, Jul 1, 2010 at 17:14,  <gri@golang.org> wrote:
>
> http://codereview.appspot.com/1687047/diff/17001/18001
> File doc/go_spec.html (right):
>
> http://codereview.appspot.com/1687047/diff/17001/18001#newcode4392
> doc/go_spec.html:4392: denoting an array.
> I thought we also allowed pointers?
>
> http://codereview.appspot.com/1687047/show
>

[gri, 2010-07-02 00:21:46]

LGTM

http://codereview.appspot.com/1687047/diff/25001/26001
File doc/go_spec.html (right):

http://codereview.appspot.com/1687047/diff/25001/26001#newcode4391
doc/go_spec.html:4391: denoting an array or pointer to array.
Should one say something like:

(even if the pointer is nil).

to make it very clear?

[rsc, 2010-07-02 00:34:25]

> Should one say something like:
>
> (even if the pointer is nil).
>
> to make it very clear?

i don't have a good way to word it.
also i think it's separate from constants.
len(f()) doesn't crash if f() returns a nil *[5]int either.
there's a recent issue on the issue tracker
about being very clear about when len and
cap can be used on which nil values.
let's leave it for that.

waiting for r and the release before submitting.

[r2, 2010-07-02 00:36:57]

LGTM

Maybe add to last paragraph:

Otherwise invocations of len and cap are not constant.

[rsc, 2010-07-02 00:49:50]

*** Submitted as http://code.google.com/p/go/source/detail?r=932ad7138e12 ***

spec: restrict when len(x) is constant

R=gri, iant, ken2, r
CC=golang-dev
http://codereview.appspot.com/1687047