Subject: Re: [boost] [AFIO] Formal review
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2015-09-02 19:05:32
On 2 Sep 2015 at 15:44, Jeremy Maitin-Shepard wrote:
> > The approach you just suggested doesn't handle the case where you
> > open a file, and then it gets moved before the sibling lookup.
> Fair enough; I think the desired semantics would depend on the particular
> case. It seems there is always a time at which you issue the open call in
> order to open/create the sibling file, and there will inevitably be a race
> between that and concurrently moving the original file.
Indeed. AFIO generally provides race guarantees for sibling lookups
to the level of the parent directory only i.e. the parent directory
and anything above it may be permuting and that is not racy, but the
contents of the parent directory and deeper are not race free. Each
operation has different guarantees on different platforms, and those
are listed per operation in the reference docs.
-- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk