bug-coreutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: proposed fix for some race conditions in mkdir and install


From: Paul Eggert
Subject: Re: proposed fix for some race conditions in mkdir and install
Date: Fri, 15 Sep 2006 22:24:17 -0700
User-agent: Gnus/5.1008 (Gnus v5.10.8) Emacs/21.4 (gnu/linux)

Eric Blake <address@hidden> writes:

> Would it make sense to depend on the fatal-signal or wait-process module,
> so that you can kill the child process when the parent is sent a catchable
> fatal signal?

I'm inclined not to do that, because it adds complexity for limited
benefit, and the complexity will subtract from reliability in other
ways.  I suppose it might be different if a child process were spun
off in the normal case, but child processes will be quite rare in
practice.

For what it's worth, POSIX allows this behavior, as it doesn't specify
whether utilities are uniprocess.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]