Wikimedia Bugzilla is closed!

I am able to view the captions provided in mwEmbed's timed text examples, but the showTextInterface() box hangs on "loading ..." when I supply a <mediaSource ... src="http://.../captions.xml"> CMML track in the file indicated by <video roe="http://.../mwEmbed_roe.xml">.  Both (static) XML files are valid, well-formatted, have content-type text/xml, and only two tracks are present.

I am able to trace the problem using Firebug's console:

load timed text from roe: http://.../mwEmbed_roe.xml
row data ready (no roe request)
mv_txt_load_:SHOW mv_txt_load_
video/ogg is not suported timed text fromat 
adding Track: Ht_en to v0
do load timed text: Ht_en
textCMML: loading track: undefined

pcurl.queryKey.t is undefined
        load()("source", function())
        loadAndDisplay()("Ht_en")
        (?)()(1, Object)
        each()([Object, Object 0=Object 1=Object], function(), undefined)
        getParseTimedText_rowReady()()
        getTimedTextTracks()()
        init()(div#v0)
        mvTextInterface()(div#v0)
        (?)()()
        doLoad()(undefined, undefined)
        mv_embed.js()
          var req_time = pcurl.queryKey['t'].split('/'); 
          ^-->JS Error from line 361 in libTimedText/mvTextInterface.js

native:onended:
base:onClipDone
mvEmbed:stop:v0


I believe that mvTextInterface.js is re-parsing the original mwEmbed_roe.xml, then retrieving the first <mediaSource> element (the video itself) rather than the second <mediaSource> listed under <track id="t" provides="text layers">.  I am not aware of a solution or workaround, or for that matter how my test case differs from the provided examples.

Any ideas?
-George


Minefield 3.6a1pre build 20090416052000
mvEmbed SVN 50996