Hi, > > the traceback below. I'm using AuthKit which may be the "unicode" > > source. > > Don't be irritated by the "Panda3d" references - it's a > normal Python > > 2.4 version below it. > > > > File > > "C:\Panda3D-1.3.2\python\lib\site-packages\paste-1.3-py2.4.egg\paste > > \auth\au > > th_tkt.py", line 163, in calculate_digest > > digest0 = md5.new(encode_ip_timestamp(ip, timestamp) + secret > > + userid > > + '\0' > > UnicodeDecodeError: 'ascii' codec can't decode byte 0xbb in > position > > 5: > > ordinal not in range(128) > > It's actually from the userid being unicode data I believe. > I've seen others report this error before as well. If userid > is unicode, its going to try and make them all unicode > objects for the + I believe, which then causes the error.
After I changed the timestamp manipulation (shifting bits etc) and the Chr() to unichr() coding the error disappeared. So everything is now based on unicode which seems fine for md5. I don't know why the ip/timestamp encoding must be done in this way - is there a security risk if a different method is used? Regards, Andrew _______________________________________________ Paste-users mailing list [email protected] http://webwareforpython.org/cgi-bin/mailman/listinfo/paste-users
