pytz: Why is normalize needed when converting between timezones?

I'm reading the not so complete pytz documentation and I'm stuck on understand one part of it.

Converting between timezones also needs special attention. This also needs to use the normalize method to ensure the conversion is correct.

>>> utc_dt = utc.localize(datetime.utcfromtimestamp(1143408899))
>>> utc_dt.strftime(fmt)
'2006-03-26 21:34:59 UTC+0000'
>>> au_tz = timezone('Australia/Sydney')
>>> au_dt = au_tz.normalize(utc_dt.astimezone(au_tz))
>>> au_dt.strftime(fmt)
'2006-03-27 08:34:59 EST+1100'
>>> utc_dt2 = utc.normalize(au_dt.astimezone(utc))
>>> utc_dt2.strftime(fmt)
'2006-03-26 21:34:59 UTC+0000'

I tried this very example without using normalize and it turned out just the same. In my opinion this example doesn't really explain why we have to use normalize when converting between datetime objects in different timezones.

Would someone please give me an example (like the one above) where the result differs when not using normalize.

Thanks

Solution

From the pytz documentation:

In addition, if you perform date arithmetic on local times that cross DST boundaries, the results may be in an incorrect timezone (ie. subtract 1 minute from 2002-10-27 1:00 EST and you get 2002-10-27 0:59 EST instead of the correct 2002-10-27 1:59 EDT). A normalize() method is provided to correct this. Unfortunately these issues cannot be resolved without modifying the Python datetime implementation.