code review 7415050: spec: terminating statements for functions

spec: terminating statements for functions

The only functional change is the new section
on terminating statements.

There is a minor syntax rewrite (not change)
of function declarations to make it easier to
refer to the notion of a function from all places
where it is used (function decls, method decls,
and function literals).

Includes some minor fixes/additions of missing links.

Based closely on Russ' proposal.

Fixes issue 65.

[gri, 2013-03-04 04:41:14]

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

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

[bradfitz, 2013-03-04 06:06:32]

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3887
doc/go_spec.html:3887: <li>both branches are terminating statements.</li>
s/both/all/ ?

[gri, 2013-03-04 17:31:41]

Waiting for comments from more people.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3887
doc/go_spec.html:3887: <li>both branches are terminating statements.</li>
On 2013/03/04 06:06:33, bradfitz wrote:
> s/both/all/ ?

There are exactly two branches (at the way this rule is formulated), so I wanted
to highlight that. But not feeling strongly.

[bradfitz, 2013-03-04 17:34:22]

On Mon, Mar 4, 2013 at 9:31 AM, <gri@golang.org> wrote:

> Waiting for comments from more people.
>
>
>
>
>
https://codereview.appspot.**com/7415050/diff/13001/doc/go_**spec.html<https:...
> File doc/go_spec.html (right):
>
> https://codereview.appspot.**com/7415050/diff/13001/doc/go_**
>
spec.html#newcode3887<https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3887>
> doc/go_spec.html:3887: <li>both branches are terminating
> statements.</li>
> On 2013/03/04 06:06:33, bradfitz wrote:
>
>> s/both/all/ ?
>>
>
> There are exactly two branches (at the way this rule is formulated), so
> I wanted to highlight that. But not feeling strongly.
>

yeah, I realized that after I mailed that.  "both" is better.

[rsc, 2013-03-04 18:06:28]

LGTM

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3929
doc/go_spec.html:3929: Flow of control does not continue past a terminating
statement.
This is true, but it doesn't seem to be used anywhere. I think it might be too
easy to misread as saying the reverse, namely that if flow of control does not
continue past a statement, then it is a terminating statement (not true), so I'm
inclined to drop this.

[gri, 2013-03-04 18:27:23]

Any other takers?

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3929
doc/go_spec.html:3929: Flow of control does not continue past a terminating
statement.
On 2013/03/04 18:06:28, rsc wrote:
> This is true, but it doesn't seem to be used anywhere. I think it might be too
> easy to misread as saying the reverse, namely that if flow of control does not
> continue past a statement, then it is a terminating statement (not true), so
I'm
> inclined to drop this.

Done.

[r, 2013-03-04 18:35:54]

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode1472
doc/go_spec.html:1472: A <i>block</i> is a (possibly empty) sequence of
declarations and statements
drop the parens.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3878
doc/go_spec.html:3878: A <a href="#Blocks">block</a> in which the statement list
ends in a terminating statement.
"ends" isn't well-defined.
A block with a non-empty statement list in which the final statement of the list
is a terminating statement.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3887
doc/go_spec.html:3887: <li>both branches are terminating statements.</li>
it's a bit subtle because of 'if .. else if .. else if .. else if ..', but i
believe 'both' is correct.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3895
doc/go_spec.html:3895: <li>there is no explicit loop condition.</li>
"no" isn't quite the same as empty. also there's this:

for true { }

which the rule does not (and should not) define as a terminating statement.

i think you want to say

the loop condition is missing or empty

which is very clear.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3905
doc/go_spec.html:3905: or <a href="#Fallthrough_statements">"fallthrough"
statement</a>.</li>
possibly labeled fallthrough statement

[gri, 2013-03-04 19:35:00]

PTAL

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode1472
doc/go_spec.html:1472: A <i>block</i> is a (possibly empty) sequence of
declarations and statements
On 2013/03/04 18:35:54, r wrote:
> drop the parens.

Done.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3878
doc/go_spec.html:3878: A <a href="#Blocks">block</a> in which the statement list
ends in a terminating statement.
On 2013/03/04 18:35:54, r wrote:
> "ends" isn't well-defined.
> A block with a non-empty statement list in which the final statement of the
list
> is a terminating statement.

Added extra sentence defining what "ends" means instead since it's used not just
here but below as well.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3887
doc/go_spec.html:3887: <li>both branches are terminating statements.</li>
On 2013/03/04 18:35:54, r wrote:
> it's a bit subtle because of 'if .. else if .. else if .. else if ..', but i
> believe 'both' is correct.

yes. The if-syntax has two branches.

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3895
doc/go_spec.html:3895: <li>there is no explicit loop condition.</li>
On 2013/03/04 18:35:54, r wrote:
> "no" isn't quite the same as empty. also there's this:
> 
> for true { }
> 
> which the rule does not (and should not) define as a terminating statement.
> 
> i think you want to say
> 
> the loop condition is missing or empty
> 
> which is very clear.

Don't quite agree. "true" is an explicit loop condition. "Missing or empty"
implies that it can be missing, or it can be empty, and I don't know what
"empty" means if not "missing". Rephrased to:

"the loop condition is absent."

This is the wording used in the spec for "for".

https://codereview.appspot.com/7415050/diff/13001/doc/go_spec.html#newcode3905
doc/go_spec.html:3905: or <a href="#Fallthrough_statements">"fallthrough"
statement</a>.</li>
On 2013/03/04 18:35:54, r wrote:
> possibly labeled fallthrough statement

Done.

[rsc, 2013-03-04 19:56:12]

LGTM

Still looks good to me, but wait for Rob.

https://codereview.appspot.com/7415050/diff/23001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/23001/doc/go_spec.html#newcode3931
doc/go_spec.html:3931: is not empty and if its final statement is terminating.
s/if //

[gri, 2013-03-04 20:26:15]

PTAL

https://codereview.appspot.com/7415050/diff/23001/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/23001/doc/go_spec.html#newcode3931
doc/go_spec.html:3931: is not empty and if its final statement is terminating.
On 2013/03/04 19:56:12, rsc wrote:
> s/if //

Done.

[r, 2013-03-04 20:41:05]

LGTM

[iant, 2013-03-04 20:55:26]

LGTM

https://codereview.appspot.com/7415050/diff/13002/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13002/doc/go_spec.html#newcode3859
doc/go_spec.html:3859: A terminating statement is one the the following:
s/the the/of the/

[gri, 2013-03-04 20:58:23]

https://codereview.appspot.com/7415050/diff/13002/doc/go_spec.html
File doc/go_spec.html (right):

https://codereview.appspot.com/7415050/diff/13002/doc/go_spec.html#newcode3859
doc/go_spec.html:3859: A terminating statement is one the the following:
On 2013/03/04 20:55:26, iant wrote:
> s/the the/of the/

wow.

fixed.

[rsc, 2013-03-04 21:17:47]

LGTM

[gri, 2013-03-04 21:55:54]

*** Submitted as https://code.google.com/p/go/source/detail?r=763763e16eb3 ***

spec: terminating statements for functions

The only functional change is the new section
on terminating statements.

There is a minor syntax rewrite (not change)
of function declarations to make it easier to
refer to the notion of a function from all places
where it is used (function decls, method decls,
and function literals).

Includes some minor fixes/additions of missing links.

Based closely on Russ' proposal.

Fixes issue 65.

R=rsc, r, iant, ken, bradfitz
CC=golang-dev
https://codereview.appspot.com/7415050