Bug 144145 – [JHBuild] Move to upstream OpenWebRTC

• Component: Platform
• Status: RESOLVED FIXED

[Philippe Normand]

.

[Philippe Normand]

Created attachment 251546 [details]
patch

[Philippe Normand]

Committed r183388: <http://trac.webkit.org/changeset/183388>

[Martin Robinson]

This is failing on the bots:

*** Checking out openwebrtc *** [32/33]
git remote set-url origin /home/slave/webkitgtk/jhbuild-mirror/openwebrtc.git
git remote update origin
Fetching origin
git fetch
git checkout 13516c7f79a0c48bb411464f7613d4b426c70f5b
fatal: reference is not a tree: 13516c7f79a0c48bb411464f7613d4b426c70f5b
*** Error during phase checkout of openwebrtc: ########## Error running git checkout

[Martin Robinson]

Locally the checkout works, but fails to build with clang:

  CC       owr_transport_agent.lo
owr_transport_agent.c:2339:14: error: use of undeclared identifier 'GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ'; did
      you mean 'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ'?
        case GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ:
             ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
             GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ
/home/martin/WebKit-old/WebKitBuild/DependenciesGTK/Root/include/gstreamer-1.0/gst/rtp/gstrtcpbuffer.h:92:3: note: 
      'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ' declared here
  GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ = 5,
  ^
1 error generated.

[Martin Robinson]

There seem to be multiple issues with this, so I'm just going to revert it. Sorry. :/

[Martin Robinson]

(In reply to comment #4)
> Locally the checkout works, but fails to build with clang:
> 
>   CC       owr_transport_agent.lo
> owr_transport_agent.c:2339:14: error: use of undeclared identifier
> 'GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ'; did
>       you mean 'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ'?
>         case GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ:
>              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>              GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ
> /home/martin/WebKit-old/WebKitBuild/DependenciesGTK/Root/include/gstreamer-1.
> 0/gst/rtp/gstrtcpbuffer.h:92:3: note: 
>       'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ' declared here
>   GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ = 5,
>   ^
> 1 error generated.

Zan informed me that this issue is due to a missing patch in gst-plugins-base. We should change the JHBuild version when adding patches to existing modules. I believe this will cause JHBuild to actually apply the patch. Not doing so can cause problems.

[Philippe Normand]

(In reply to comment #3)
> This is failing on the bots:
> 
> *** Checking out openwebrtc *** [32/33]
> git remote set-url origin /home/slave/webkitgtk/jhbuild-mirror/openwebrtc.git
> git remote update origin
> Fetching origin
> git fetch
> git checkout 13516c7f79a0c48bb411464f7613d4b426c70f5b
> fatal: reference is not a tree: 13516c7f79a0c48bb411464f7613d4b426c70f5b
> *** Error during phase checkout of openwebrtc: ########## Error running git
> checkout

On which bot? I lost a fair amount of time today dealing with the "jhbuild mirrors" not  handling this patch very well but I think now the buildbots are almost fine. Many tests failing but unrelated with this patch AFAIK... We need renewed gardening efforts...

[Philippe Normand]

(In reply to comment #6)
> (In reply to comment #4)
> > Locally the checkout works, but fails to build with clang:
> > 
> >   CC       owr_transport_agent.lo
> > owr_transport_agent.c:2339:14: error: use of undeclared identifier
> > 'GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ'; did
> >       you mean 'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ'?
> >         case GST_RTCP_RTPFB_TYPE_RTCP_SR_REQ:
> >              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> >              GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ
> > /home/martin/WebKit-old/WebKitBuild/DependenciesGTK/Root/include/gstreamer-1.
> > 0/gst/rtp/gstrtcpbuffer.h:92:3: note: 
> >       'GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ' declared here
> >   GST_RTCP_RTPFB_TYPE_RCTP_SR_REQ = 5,
> >   ^
> > 1 error generated.
> 
> Zan informed me that this issue is due to a missing patch in
> gst-plugins-base. We should change the JHBuild version when adding patches
> to existing modules. I believe this will cause JHBuild to actually apply the
> patch. Not doing so can cause problems.

I didn't know about this other JHBuild peculiarity. I thought the WebKitBuild/DependenciesGTK was wiped out after each jhbuild.modules update anyway?

[Martin Robinson]

(In reply to comment #7)

> On which bot? I lost a fair amount of time today dealing with the "jhbuild
> mirrors" not  handling this patch very well but I think now the buildbots
> are almost fine. Many tests failing but unrelated with this patch AFAIK...
> We need renewed gardening efforts...

The bots are looking better now. :)

(In reply to comment #8)

> I didn't know about this other JHBuild peculiarity. I thought the WebKitBuild/DependenciesGTK was wiped out after each jhbuild.modules update anyway?

They are, but I think to work around small issues like these. If someone runs jhbuild-wrapper manually though (which I do sometimes), properly versioning can prevent problems.