The HTML formatter currently fails to output debug if it contains characters outside of the default encoding. The page source in B2G often contains such characters.

This is how we fixed this in the HTML report currently generated by the Marionette test runner. I'm open to other suggestions.

Comment on attachment 8509485 [details] [diff] [review] Fix encoding to utf-8. v1.0 Review of attachment 8509485 [details] [diff] [review]: ----------------------------------------------------------------- ::: testing/mozbase/mozlog/mozlog/structured/formatters/html/html.py @@ +96,5 @@ > > def make_result_html(self, data): > + # change default encoding to avoid encoding problem for page source > + reload(sys) > + sys.setdefaultencoding('utf-8') This is terrifying ;) I think we should try to track down the places that are assuming ascii and actually fox them to be explicit. If you send me a log file that breaks I'll help with this.

(In reply to James Graham [:jgraham] from comment #2) > I think we should try to track down the places that are assuming ascii and > actually fox them to be explicit. If you send me a log file that breaks I'll > help with this. I was hoping you'd say that. Traceback and page source that caused it coming up!

It looks like the content argument in base64.b64encode(content) must be a unicode string. If you instead pass in content.encode("utf8") I expect it will work.

(In reply to James Graham [:jgraham] from comment #6) > It looks like the content argument in base64.b64encode(content) must be a > unicode string. If you instead pass in content.encode("utf8") I expect it > will work. Yep, that works well. Thanks!