For the upcoming update tests for Aurora 6.0a2 builds we have to support the new 'auroratest' channel. I will work on it right away.

Last qref was broken. Now the real patch.

We missed to add the supported aurora channels to the testrun_all.py script. This is now included.

Landed as: http://hg.mozilla.org/qa/mozmill-automation/rev/14e8e8a7d5b4

This has caused a regression in our release_update.py testrun execution: "UpdateTestRun" object has no attribute _mozmill

This fixes the broken handling of the channel inside the channel-prefs.js file. I talked to Jeff and for the future we will have a utility function to read/write pref files. For now this workaround has to work.

I missed to change group(1) to group(0). With the given regex both will return the same value, but with this change we will be consistent in each usage. Got r+ from Jeff on IRC.

Follow-up fix landed as: http://hg.mozilla.org/qa/mozmill-automation/rev/3363853a5dd4

You could also eval in a sandbox, taking advantage that JS and python have similar syntax. Here is a naive implementation: import re comment = re.compile('/\*([^*]|[\r\n]|(\*+([^*/]|[\r\n])))*\*+/', re.MULTILINE) def read(filename): token = '##//' # magical token lines = [i.strip() for i in file(filename).readlines() if i.strip()] retval = {} _lines = [] for line in lines: if line.startswith('#'): continue if '//' in line: line = line.replace('//', token) _lines.append(line) string = '\n'.join(_lines) string = re.sub(comment, '', string) def pref(a, b): retval[a] = b lines = [i.strip().rstrip(';') for i in string.split('\n') if i.strip()] for line in lines: try: _globals = {'retval': retval, 'pref': pref, 'user_pref': pref, 'true': True, 'false': False} eval(line, _globals, {}) except SyntaxError: print line import pdb; pdb.set_trace() for key in retval: if isinstance(retval[key], basestring) and token in retval[key]: retval[key] = retval[key].replace(token, '//') return retval