code review 13042044: math/big: fix nil bug in GobEncode

math/big: fix nil bug in GobEncode

Update issue 5305.
This handles the case where the nil pointers are inside a slice.
A top-level nil pointer is harder, maybe fundamentally broken by gob's model.
Thinking required.
However, a slice is the important case since people don't expect to be sending
top-level nils much, but they can arise easily in slices.

[r, 2013-08-16 06:50:34]

Hello golang-dev@googlegroups.com,

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

[josharian, 2013-08-16 19:39:42]

On 2013/08/16 06:50:34, r wrote:
> Hello mailto:golang-dev@googlegroups.com,
> 
> I'd like you to review this change to
> https://code.google.com/p/go/

This fixes the panic, but an encode/decode round trip will still fail, with
error "Int.GobDecode: no data". I'm not sure whether this an acceptable outcome.
(I still don't really understand how gob is supposed to work with nil pointers,
despite reading up on it.)

In case it helps, here's a test for math/big/int_test.go that exhibits the
decode failure:

// https://code.google.com/p/go/issues/detail?id=5305
func TestGobEncodingNilIntPointerSlice(t *testing.T) {
	buf := new(bytes.Buffer)
	enc := gob.NewEncoder(buf)
	dec := gob.NewDecoder(buf)

	in := struct{ IntPtrs []*Int }{make([]*Int, 3)}
	err := enc.Encode(&in)
	if err != nil {
		t.Errorf("gob encode failed: %q", err)
	}
	out := struct{ IntPtrs []*Int }{make([]*Int, 3)}
	err = dec.Decode(&out)
	if err != nil {
		t.Errorf("gob decode failed: %q", err)
	}
	// not sure whether this next check is supposed to pass; gobs and nils still
confuse me
	if !reflect.DeepEqual(in, out) {
		t.Errorf("gob round trip failed, want %q got %q", in, out)
	}
}

[r, 2013-08-19 00:50:58]

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

Please take another look.

[adg, 2013-08-19 01:11:09]

LGTM

[r, 2013-08-19 01:22:16]

*** Submitted as https://code.google.com/p/go/source/detail?r=b20a8e563992 ***

math/big: fix nil bug in GobEncode

Update issue 5305.
This handles the case where the nil pointers are inside a slice.
A top-level nil pointer is harder, maybe fundamentally broken by gob's model.
Thinking required.
However, a slice is the important case since people don't expect to be sending
top-level nils much, but they can arise easily in slices.

R=golang-dev, josharian, adg
CC=golang-dev
https://codereview.appspot.com/13042044

[josharian, 2013-08-19 16:33:57]

On 2013/08/19 01:22:16, r wrote:
> *** Submitted as https://code.google.com/p/go/source/detail?r=b20a8e563992 ***
> 
> math/big: fix nil bug in GobEncode
> 
> Update issue 5305.
> This handles the case where the nil pointers are inside a slice.
> A top-level nil pointer is harder, maybe fundamentally broken by gob's model.
> Thinking required.
> However, a slice is the important case since people don't expect to be sending
> top-level nils much, but they can arise easily in slices.
> 
> R=golang-dev, josharian, adg
> CC=golang-dev
> https://codereview.appspot.com/13042044

LGTM