According to the documentation:
"The fill character can be any character other than ‘}’ (which signifies the end of the field)."
http://docs.python.org/dev/library/string.html#format-specification-mini-language


However the format() builtin accepts both '{' and '}' characters:

>>> format(42, '}^6')
'}}42}}'

>>> format(42, '{^6')
'{{42{{'


And the string method rejects both characters.

>>> '{:}^6}'.format(42)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: Single '}' encountered in format string

>>> '{:{^6}'.format(42)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: unmatched '{' in format

Looks more like a behavior bug than a doc bug to me. Does the doc comply with the PEP?

The PEP 3101 does not prohibit any character for the 'fill' argument.

Another example which just works:

>>> '{:{fill}^6}'.format(42, fill='{')
'{{42{{'

>>> '{:{fill}^6}'.format(42, fill='}')
'}}42}}'


I don't care if '{' and '}' are prohibited when using simple formatting syntax.  This is not a common use case, and there are workarounds (either using format() builtin or the recursive formatting).

A documentation fix could be enough.

Thanks for clarifying.

Fixed docs in r84553.

(That builtin format() supports this is no surprise, and has no influence on the validity in format strings.)

Sorry to respond late.

The reason for this is that the parsing of the string (as delimited by "{" and "}") happens before the results are then interpreted as format specifiers. There's no way around it, short of the parser understanding every object's formatting language, which is of course not possible. It could be special cased for string, int, and float format specifiers, but that doesn't make much sense.

I think the doc change is good.