code review 157082: Permit omission of hi bound in slices.

Permit omission of hi bound in slices.

[gri, 2009-11-19 03:55:08]

Hello r,

I'd like you to review the following change.

[r, 2009-11-19 04:01:43]

seems ok
the spacing looks odd  [a+b :]  should be [a+b:] ?
hold off until compilers have it?

[rsc, 2009-11-19 04:41:24]

now that i see it, i think i agree: x[a+b:] is better

[rsc, 2009-11-19 22:47:21]

http://codereview.appspot.com/157082/diff/30/1003
File src/pkg/go/printer/nodes.go (right):

http://codereview.appspot.com/157082/diff/30/1003#newcode668
src/pkg/go/printer/nodes.go:668: // TODO(gri): should treat[] like parentheses
and undo one level of depth
I tried that on the index-heavy expressions
in the test list (all the index+shift+or expressions)
at one point but it didn't look right.

http://codereview.appspot.com/157082/diff/30/1003#newcode672
src/pkg/go/printer/nodes.go:672: if x.Colon.IsValid() {
(Assuming this is Colon.Pos.IsValid underneath)

I'm not sure about using x.Colon.IsValid() as the way to
tell whether this is a slice.  Programs that generate AST
might not include position information, in which case
this test will fail.  It will be a weird failure.

Maybe it makes sense to have a separate AST type for
slicing vs indexing?  Or Colon bool instead of Colon Pos?

[rsc, 2009-11-20 00:16:28]

http://codereview.appspot.com/157082/diff/47/50
File src/pkg/go/ast/ast.go (right):

http://codereview.appspot.com/157082/diff/47/50#newcode183
src/pkg/go/ast/ast.go:183: IndexExpr;
Seems awkward to me: why not just put X and Index here.

http://codereview.appspot.com/157082/diff/47/51
File src/pkg/go/ast/walk.go (right):

http://codereview.appspot.com/157082/diff/47/51#newcode134
src/pkg/go/ast/walk.go:134: Walk(v, &n.IndexExpr);
I'm not sure about putting IndexExpr inside SliceExpr.
This demonstrates why: a SliceExpr is not (except syntactically)
just an IndexExpr and an extra piece, any more than a 
BinaryExpr is a UnaryExpr with an extra piece.

A walk function looking for actual IndexExprs probably
doesn't want to see half-SliceExprs.

[rsc, 2009-11-20 00:38:03]

LGTM

[gri, 2009-11-20 00:42:07]

*** Submitted as http://code.google.com/p/go/source/detail?r=63f563584f4c ***

Permit omission of hi bound in slices.

R=r, rsc
http://codereview.appspot.com/157082