Issue 2717: tempfile.mkstempf - Python tracker (original) (raw)

Created on 2008-04-29 02:11 by skip.montanaro, last changed 2022-04-11 14:56 by admin. This issue is now closed.

Messages (7)
msg65951 - (view)	Author: Skip Montanaro (skip.montanaro) *	Date: 2008-04-29 02:11
On python-dev Guido lamented the fact that tempfile.mkstemp() returns a file descriptor instead of a file object. This patch adds tempfile.mkstempf to remedy that.
msg65961 - (view)	Author: André Malo (ndparker) *	Date: 2008-04-29 11:25
It should catch OSError on fdopen, close fd and reraise, I think.
msg65964 - (view)	Author: Skip Montanaro (skip.montanaro) *	Date: 2008-04-29 11:50
Thanks. New patch attached.
msg65965 - (view)	Author: Alyssa Coghlan (ncoghlan) *	Date: 2008-04-29 12:24
What does this offer over using NamedTemporaryFile(delete=False)?
msg65966 - (view)	Author: Skip Montanaro (skip.montanaro) *	Date: 2008-04-29 13:54
Nick> What does this offer over using NamedTemporaryFile(delete=False)? Nothing I suppose. I wasn't aware of the similarity of the two (I always use mkstemp()) or the delete arg to NamedTemporaryFile, and assumed Guido's complaint about mkstemp's shortcoming indicated that the functionality he desired didn't exist. Skip
msg65972 - (view)	Author: Alyssa Coghlan (ncoghlan) *	Date: 2008-04-29 15:40
The delete flag on NamedTemporaryFile is new for 2.6, so it isn't surprising people aren't familiar with it. The other advantage of using NamedTemporaryFile is that its name attribute is set correctly, which isn't the case when using mkstemp directly.
msg66392 - (view)	Author: Skip Montanaro (skip.montanaro) *	Date: 2008-05-08 02:28
The conclusion on python-dev is that NamedTemporaryFile is an adequate replacement for mktemp, so this isn't needed.