210337 - Remove dead cruft from XPCOM

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Benjamin Smedberg]

XPCOM has some dead code that can be removed. Some of it still appears to be
#included by not actually used. So I've created a patch that removes unnecessary
#includes and stops building cruft.

The following files will be removed (I didn't include them in the patch, for
brevity):
xpcom/ds/nsHashtableEnumerator.h
xpcom/ds/nsHashtableEnumerator.cpp
xpcom/ds/nsIArena.idl
xpcom/ds/nsArena.h
xpcom/ds/nsArena.cpp

Comment 1

[Benjamin Smedberg]

Attachment: remove XPCOM cruft

Comment 2

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 126277 [details] [diff] [review]
remove XPCOM cruft

r+sr=me on the content/* parts of this patch.

Comment 3

[Alec Flett]

Comment on attachment 126277 [details] [diff] [review]
remove XPCOM cruft

-NS_COM nsresult NS_NewObserver(nsIObserver** anObserver, nsISupports* outer)
-{
-    return nsObserver::Create(outer, NS_GET_IID(nsIObserver),
(void**)anObserver);
-}
-

can we get rid of nsObserver::Create()? actually why do we need nsObserver at
all? what does a default observer even do? (!)

r/sr=alecf on the xpcom stuff... even the removal of NS_NewISupportsPRBool.. :(

Comment 4

[Benjamin Smedberg]

alec, the only reason I didn't remove nsObserver altogether is because it is a frozen interface. I didn't (and don't) know the rules about removing contractID implementations of frozen interfaces.

Really I think we need to be much more clear about which contractIDs are frozen, and what freezing them really means... (this  contractID will always support such and such set of interfaces). The header files that mark the contractIDs should have /* @status FROZEN */ comments just like the interface files do.

Comment 5

[Doug Turner (:dougt)]

NS_NewObserver isn't an interface and it is not frozen. 

Comment 6

[Doug Turner (:dougt)]

what alec said.  nsObserver.* is bogus.

Comment 7

[Benjamin Smedberg]

fixed