784828 - SVG Code to query/insert into nsSVGAttrTearoffTable should use nsRefPtr instead of explicit NS_ADDREF

Status: RESOLVED FIXED

(Core :: SVG, defect)

Description

[Daniel Holbert [:dholbert]]

Currently, we've got a code-pattern across our "look up or insert into nsSVGAttrTearoffTable" code that looks like this:

55 /* static */ already_AddRefed<DOMSVGAnimatedNumberList>
56 DOMSVGAnimatedNumberList::GetDOMWrapper(SVGAnimatedNumberList *aList,
57                                         nsSVGElement *aElement,
58                                         uint8_t aAttrEnum)
59 {
60   DOMSVGAnimatedNumberList *wrapper =
61     sSVGAnimatedNumberListTearoffTable.GetTearoff(aList);
62   if (!wrapper) {
63     wrapper = new DOMSVGAnimatedNumberList(aElement, aAttrEnum);
64     sSVGAnimatedNumberListTearoffTable.AddTearoff(aList, wrapper);
65   }
66   NS_ADDREF(wrapper);
67   return wrapper;
68 }
https://mxr.mozilla.org/mozilla-central/source/content/svg/content/src/DOMSVGAnimatedNumberList.cpp#55

The manual NS_ADDREF boilerplate there is a bit boilerplate-ish and scary.  I think we should clean these up to use nsRefPtr for the refcounting, and use .forget() to return our (type-safe) already_AddRefed return-value.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: patch 1: DOMSVG*List

This patch fixes all instances of this pattern in DOMSVG*List.cpp.

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: patch 2: nsSVGLength2

...and this one fixes nsSVGLength2.

The patch also does the following:
 (a) It makes the tearoff tables in nsSVGLength2 store their actual concrete types, instead of nsIDOMWhatever types.  This matches the tearoff tables elsewhere, and it lets us use nsRefPtr<ConcreteClass> when we query the table.

 (b) It makes the DOMBaseVal and DOMAnimVal structs public, since that's required for (a) to compile, and there's no strong reason for them to be private. (and the similar helper-class DOMAnimatedLength is already public)

Comment 3

[Daniel Holbert [:dholbert]]

Try run: https://tbpl.mozilla.org/?tree=Try&rev=d4958b0e9fda

Comment 4

[Daniel Holbert [:dholbert]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/165cd64fb08a
https://hg.mozilla.org/integration/mozilla-inbound/rev/c80863b9bf12

Comment 5

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/165cd64fb08a
https://hg.mozilla.org/mozilla-central/rev/c80863b9bf12