These are the changeset details:
changeset:   18337:c2a60de91d2c
branch:      legacy-trunk
user:        Guido van Rossum <guido@python.org>
date:        Fri Jun 29 13:06:06 2001 +0000
summary:     Solve SF bug #231249: cgi.py opens too many (temporary) files.


You're right that we might use environment variables or tempfile.tempdir
to attain the same goal but this would impact _all_ of the code executed
during request data parsing given the monolithic construction of
FieldStorage. This implies that the context of every call to tempfile
members would be impacted during this process.  Presently this is not a
problem at all, but this looks fragile for future developments. 

On the other hand:
1) this has the advantage of not changing FieldStorage interface,
2) this alleviates us of wondering if passing to FieldStorage
constructor all of the arguments to NamedTemporaryFile is a possibility
worth considering ;-).

After pondering this for a while I think the simpler is the better and I
propose to add documentation to inform the reader that changing the
temp directory through os.environ of tempfile.tempdir might worth
consideration.

As for other use cases for changing the temp directory, I thought about
letting the user choose the FS of its choice with regard to
performance or security (crypted FS) or even having the temp files created in a directory with 700 permissions. 

Last, you're perfectly right about the None argument.

I fiddled last night with setting an environment to deploy and test a patched Python (I had some problem to understand what happened when I encountered 6755). This now works and the patch does not introduced any regression. I still have to add unit tests (I only tested with my embryonic cgi script) and update the documentation before to send the patch. I should be able to do that in a few days at most.