769801 - Signaling code needs separate build for linking with C++ tests.

Status: RESOLVED DUPLICATE of bug 792188

(Core :: WebRTC: Signaling, defect)

Description

[Ethan Hugg [:ehugg]]

Signaling code needs to be built twice.  Once for inclusion is the browser and differently for the test harness.  In particular MOZILLA_INTERNAL_API should only  be defined in the first case.

Comment 1

[Ethan Hugg [:ehugg]]

Attachment: Build signaling twice for FF and Test

Comment 2

[Ethan Hugg [:ehugg]]

Here's the patch that should make signaling twice.

The makes differ by the define of MOZILLA_INTERNAL_API:
ehugg@ehugg-linux:~/mozilla/alder/alder/obj-x86_64-unknown-linux-gnu/media/webrtc$ grep INTERNAL signaling/signaling_ecc/Makefile 
  -DMOZILLA_INTERNAL_API \
  -DMOZILLA_INTERNAL_API \
ehugg@ehugg-linux:~/mozilla/alder/alder/obj-x86_64-unknown-linux-gnu/media/webrtc$ grep INTERNAL signalingtest/signaling_ecc/Makefile 
ehugg@ehugg-linux:~/mozilla/alder/alder/obj-x86_64-unknown-linux-gnu/media/webrtc$ 

Which makes the objs differ like this:
ehugg@ehugg-linux:~/mozilla/alder/alder/obj-x86_64-unknown-linux-gnu/media/webrtc$ nm signaling/signaling_ecc/src/peerconnection/PeerConnectionImpl.o | grep NS_DebugBreak
                 U NS_DebugBreak_P
ehugg@ehugg-linux:~/mozilla/alder/alder/obj-x86_64-unknown-linux-gnu/media/webrtc$ nm signalingtest/signaling_ecc/src/peerconnection/PeerConnectionImpl.o | grep NS_DebugBreak
                 U NS_DebugBreak


f=EKR - to check if this give what we need
f=RJesup - to check if this is the best way to do this.

Comment 3

[Randell Jesup [:jesup] (needinfo me)]

Comment on attachment 638021 [details] [diff] [review]
Build signaling twice for FF and Test

+        ['build_for_test==1', {
+          'defines' : [
+          ],
+        }], 

This doesn't do anything; remove.

Comment 4

[Eric Rescorla (:ekr)]

Funny you should mention that, since I just added a clause here.

Comment 5

[Ethan Hugg [:ehugg]]

This was put into M-C as part of the Alder merge - Bug 792188