Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!news.glorb.com!newsfeed.xs4all.nl!newsfeed2a.news.xs4all.nl!xs4all!newsgate.cistron.nl!newsgate.news.xs4all.nl!post.news.xs4all.nl!not-for-mail
Date: Thu, 13 Mar 2014 13:49:27 +1100
From: Cameron Simpson <cs@zip.com.au>
To: python-list@python.org
Subject: Re: Save to a file, but avoid overwriting an existing file
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <85zjkvrlpc.fsf@benfinney.id.au>
User-Agent: Mutt/1.5.21 (2010-09-15)
References: <85zjkvrlpc.fsf@benfinney.id.au>
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.8121.1394678982.18130.python-list@python.org>
Lines: 36
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:68321

On 13Mar2014 11:22, Ben Finney <ben+python@benfinney.id.au> wrote:
> Cameron Simpson <cs@zip.com.au> writes:
> > Therefore you need to continue _only_ if you get EEXIST. Otherwise
> > abort.
> 
> If you target Python 3.3 or later, you can catch “FileExistsError”
> <URL:http://docs.python.org/3/library/exceptions.html#FileExistsError>
> which is far simpler than messing around with ‘errno’ values.

Yes and no.

Firstly, we Old Farts find these new ways confusing:-)

Secondly, and perhaps more importantly, if I were using the O_EXCL
et al constants in a call to os.open as the OP is doing, I would
want to be equally picky about what I recognise in the exception;
I _want_ to use error explicitly. In particular, the behaviour of
"os.open" is pretty much a bare metal call to the POSIX "open(2)"
OS interface, and so I want my response to be framed directly in
terms of the documented error responses for that API. Thus, errno
values.

Finally, if I were grouping my responses to the exception by type
and perhaps performing a few different actions depending on the
error, then the new class hierachy is a good way to do it. But for
this instance, IMO, we're catching exactly one failure mode as
"retry with another name" and reraising anything else. So I don't
find an errno test any less readable that a class comparison, and
it still feels more precise to me.

Cheers,
-- 
Cameron Simpson <cs@zip.com.au>

The road less traveled is FASTER !!
        - Luc Marcouiller, marcouil@ireq.hydro.qc.ca