TextCalendar.formatmonth is not influenced by setfirstweekday,
but the argument of constructor.

Documentation:
Depends on the first weekday as set by setfirstweekday().

actual behavior:
>>> cal0 = calendar.TextCalendar()
>>> print cal0.formatmonth(2008, 2)
   February 2008
Mo Tu We Th Fr Sa Su
             1  2  3
 4  5  6  7  8  9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29

>>> calendar.setfirstweekday(4)
>>> print cal0.formatmonth(2008, 2)
   February 2008
Mo Tu We Th Fr Sa Su
             1  2  3
 4  5  6  7  8  9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29

setfirstweekday() isn't supposed to have any influence on calendar
objects created explicitely. The function setfirstweekday() is only for
the module level interface. The documentation is wrong here. However you
*can* change the first weekday with the setfirstweekday() *method*.

Thank you for the clarification.
But then, where is the documentation about the setfirstweekday() method?

There is another wrong documentation about firstweekday. 
In the part of Calendar.iterweekdays, a method "firstweekday" appered.
But indeed it means the method "getfirstweekday()", since firstweekday
is merely an uncallable integer attribute of Calendar objects.

The doccumentation is
here:http://docs.python.org/dev/library/calendar.html#calendar.TextCalendar.formatmonth
(or in Doc/library/calendar.rst in the source).

Anyway the first of those documentation bugs is fixed now in r60649
(trunk) and r60650 (release25-maint).

You're supposed to use firstweekday as a property instead of using the
getter method getfirstweekday(). Anyway this is fixed now in r60651
(trunk) and r60652 (release25-maint)

In the message msg62100, I asked about the place of *setfirstweekday()*.
Your answer in the message msg62173 was about *formatmonth()*.
I don't think the fix is complete until some explanations are given for
firstweekday and/or its getter/setter, since they have been mentioned
from your description fixes but have no entries.