5182 - build error: HP-UX: nsRDFElement.cpp (and DOM macro)

Status: VERIFIED FIXED

(Core Graveyard :: RDF, defect, P3)

Description

[Bruce Mitchener]

Error 698: "../../../dist/./include/nsIDOMNode.h", line 125 # Error limit
reached; halting compilation.
         D    GetOwnerDocument(nsIDOMDocument** aOwnerDocument) { return
_to##GetOwnerDocument(aOwnerDocument); } \

^^^^^^^^^^^^^^^^^^^^^
NEXT gmake[3]: *** [nsRDFElement.o] Error 2

Each line like that in the macro failed.  This is visible on the HP-UX 10.20
tinderbox on the ports page (eggroll).

nsRDFElement.cpp includes nsXULTreeElement.h and on:
http://lxr.mozilla.org/seamonkey/source/rdf/content/src/nsXULTreeElement.h#43

it calls that macro, passing it mOuter->

and then compile go boom.

cc'ing vidur since the macro was defined in DOM land. assigning to waterson
since the calling of the macro is in RDF/XUL land.

Comment 1

[shaver-do-not-use-this-account]

I think the right thing to do is change:
  NS_IMETHOD    GetNodeName(nsString& aNodeName) { return
_to##GetNodeName(aNodeName); } \
to
  NS_IMETHOD    GetNodeName(nsString &aNodeName) { return
_to/**/GetNodeName(aNodeName); } \
and so forth.  I think strict C says that ## can only be used to create a valid
single token, the the latter form will at worst insert a harmless space when the
preprocessor is done with it.

Comment 2

[Chris Waterson]

Attachment: recommended patch to XPCOMGen.cpp

Comment 3

[Chris Waterson]

Vidur: I've regenerated the core DOM header files with the above patch for
idlc. Did a full rebuild on Win32: it works fine. I've also spot-checked Mac
and Linux, so it looks like this'll be okay. Give the word and I'll check in
the patch to idlc and the re-generated DOM headers.

Comment 4

[Chris Waterson]

Actually, let me correct the above statement. I regenerated _all_ the DOM
header files (all of mozilla/dom/public/idl).

Comment 5

[vidur (gone)]

Go for it, Chris.

Comment 6

[Chris Waterson]

Changes checked in.