python 2.3.1 issues DeprecationWarning for FCNTL when importing tempfile

Tim Peters tim.one at comcast.net
Sat Oct 4 17:49:06 EDT 2003


[Pierre Rouleau]
>>> Actually, i do understand that now.  The import worked because
>>> fcntl.pyc existed.  What I don't understand is why there was a file
>>> called fcntl.pyc in the directory when fcntl.py was not there (even
>>> though the source file FCNTL.py is present).

[Tim]
>> I don't know either, but am pretty sure it's not due to anything in
>> the core Python distribution.  At the far reaches of my aging
>> memory, it rings a tiny bell:  do you use any of the packages for
>> building distributions on Windows? I have the vaguest memory that
>> one of those went around compiling Python files, but didn't preserve
>> the original case of module names.  In your case, I'd ask you to
>> remember what you were doing at 10:28 on 30 Sep 2003 (that's when
>> your fcntl.pyc file got created, if we're to believe the output from
>> previous email):
>>
>> -rwxrwxrwa   1 Administrators  None      583 Sep 30 10:28 fcntl.pyc

[Pierre]
> You might be right here.  I am using the McMillan Installer and that
> would coincide. After updating to Python 2.3.1 I also update the
> Installer (got the latest 5b5_5).  However, i did not use it May 9,
> 2001 (which is the time stamp of FCNTL.py).  Unless McMillan Installer
> preserved the time stamp of a file while changing its case.

FCNTL.py on every Windows box has a 9-May-2001 timestamp -- that's just the
timestamp the Windows installer created for it when you installed Python.
That's not relevant to any of this, though, as far as I can see.  The
question remains what you were doing at 10:28 on 30 Sep 2003, because that's
the timestamp on the only file in question here (FCNTL.py would have been an
input to the compilation process that created the troublesome fcntl.pyc;
what Windows shows you is a creation-date timestamp, and compilation doesn't
change that on input files).

> I ran into other problem with the Installer (support for gettext)
> which I reported on the McMillan Bug Tracker site
> (http://www.mcmillan-inc.com/cgi-bin/BTSCGI.py/BTS/).
>
> Would you remember what version of the Installer was doing that?  I
> should probably report that to Gordon McMillan if i could reproduce
> it.

Sorry, I have only the vaguest recollection of this.  I can't even say
whether Gordon's installer-builder specifically was implicated in this aged
and feeble memory.  If you can remember what you did a few days ago, though,
you could try to provoke it again.






More information about the Python-list mailing list