317869 - CNavDtd::GetMostDerivedIID returns nsIDTD's IID

Status: RESOLVED FIXED

(Core :: DOM: HTML Parser, defect, P2)

Description

[[e-mail]]

nsExpatDriver::GetMostDerivedIID retuns nsIDTD's IID. nsViewSourceHTML::GetMostDerivedIID returns its own custom IID.
However CNavDtd::GetMostDerivedIID tries to return NS_GET_IID(CNavDtd) which happens to resolve to nsIDTD's IID. Now although I could not find any history for it, given the implementation of QueryInterface and CDTDFinder I guess that CNavDtd should have its own ID, along the lines of nsViewSourceHTML.

Comment 1

[Blake Kaplan (:mrbkap) (inactive)]

My recent cleanup accidentally removed CNavDTD's IID.

Comment 2

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: Put it back

Comment 3

[[e-mail]]

Attachment: Don't put it back

The point being that CNavDTD isn't an interface, so shouldn't have an accessor.

In fact since nobody actually calls QueryInterface for NS_INAVHTML_DTD_IID that second hunk could be changed to NS_IMPL_ISUPPORTS1(CNavDTD, nsIDTD) instead.

Comment 4

[[e-mail]]

Comment on attachment 204245 [details] [diff] [review]
Don't put it back

D'oh, I just spotted that you do need to be able to QI to the class :-[

Comment 5

[Blake Kaplan (:mrbkap) (inactive)]

The other problem is that we currently rely on GetMostDerivedIID to select the correct DTD (I'm not exactly sure why my checkin didn't break everything). I'll try to work on removing this crazyness at some point.

Comment 6

[Johnny Stenback (:jst)]

Comment on attachment 204235 [details] [diff] [review]
Put it back

r+sr=jst

Comment 7

[Blake Kaplan (:mrbkap) (inactive)]

Checked in.