199122 - IDispatch logic should not depend on order of properties

Status: VERIFIED FIXED

(Core :: XPConnect, defect, P1)

Description

[David Bradley]

The IDispatch logic should not depend on the order that a property getter/setter
methods occur during enumerating the interface.

Comment 1

[David Bradley]

Moving back, will seek an a= when reviewed. IDispatch only changes, very low risk.

Comment 2

[Adam Lock]

Attachment: Test control

This is a test control with an odd IDL interface. The interface contains
parameterized parameters, and setters and getters which are seperated rather
than being defined together.

Comment 3

[David Bradley]

*** Bug 199600 has been marked as a duplicate of this bug. ***

Comment 4

[David Bradley]

Attachment: Addresses the property method order issue

This patch actually addresses two problems. One is the property order. The
other is the parameterized property problem. The patch in bug 198668 had the
test for IsRetVal backwards, should be !IsRetVal. (Adam double check that but
it seems to be correct now). This was what was breaking in bug 199600.

There were two spots that needed to be fixed. One is the allocation of the
XPCDispInterface, as it needs to know how many unique entries there will be.
The other was the one filling out the actual member information.

I also took the liberty of cleaning up a few warnings. I took out the commented
out XPCDispParams::GetParamRef since it's not used.

Comment 5

[Adam Lock]

Comment on attachment 118994 [details] [diff] [review]
Addresses the property method order issue

It looks okay in general (incl the parameterized fix). Have you had a chance to
run it over the sample control I provided? I just want to be sure that the
setter/getter order doesn't mean the getter funcdesc gets put in mFuncDesc by
mistake (i.e. via SetTypeInfo call).

If so r=adamlock

Comment 6

[David Bradley]

Yes, it's working just fine.

Comment 7

[Alec Flett]

Comment on attachment 118994 [details] [diff] [review]
Addresses the property method order issue

sr=alecf
nice liberal use of comments! :)

Comment 8

[Alec Flett]

Comment on attachment 118994 [details] [diff] [review]
Addresses the property method order issue

sr=alecf
nice liberal use of comments! :)

Comment 9

[David Bradley]

Fix checked in

Comment 10

[Phil Schwartau]

Checkin verified -