[twill] Debugging twill

[Titus Brown]

On Wed, Nov 23, 2005 at 02:30:29PM +0530, sureshvv wrote:
-> Titus Brown wrote:
->   ->
-> > -> I was able to make it work by patching _mech_open() method in _mechanize.py as 
-> > -> follows:
-> > -> 
-> > -> Around line 281, change
-> > -> 
-> > -> <       self._response = UserAgent.open(self, self.request, data)
-> > -> 
-> > -> to
-> > -> 
-> > ->  >       try:
-> > ->  >           self._response = UserAgent.open(self, self.request, data)
-> > ->  >       except urllib2.HTTPError, e:
-> > ->  >           self._response = e
-> > -> 
-> > 
-> > It makes me more interested than ever in what is happening in your case!
-> > If you look in 'utils.py', function 'journey', I do almost exactly the
-> > same thing.  I don't have a good idea of where your approach should
-> > differ the one in twill already... I'll have to look at the code with
-> > this in mind.
-> 
-> Yep... I copied it from there :)
-> 
-> Unfortunately it is too late in journey, since mechanize already thinks that 
-> there is an error and does not save the response in self._response. Later on, 
-> when you try to viewing_html(), it says Not viewing a page.
-> 
-> So the fix needs to be done at the mechanize level.

hi, Suresh,

thanks for this report.  I finally managed to duplicate your bug -- it
depended on the restricted page being loaded *after* a non-HTML page --
and after a certain amount of agonized deliberating, decided that
patching mechanize directly is the proper way to go.

I'm a bit worried that this won't make it into the trunk mechanize
source without some more work, but that will depend entirely on what
John thinks.

thanks,
--titus