On Fri, Sep 10, 2004 at 03:15:38AM -0500, David Nicol wrote:
> I was not aware of the risk of select hanging due to a signal coming in before
> select can set its timer. Has anyone on this list ever been bitten by such a
> thing? Is Perl vulnerable to this situation or is it taken care of
> internally? The
> situation described would turn a nonblocking select call into an
> indefinately blocking
> select, but it sounds unlikely to occur in practice, and would only
> matter when none of
> the pipes are readable anyway. If one uses wait or waitpid rather
> than redefining
> $SIG{CHLD} is one vulnerable?
There are better mailing lists for those kinds of issues.
> Should I set up an asynchronous DBI mailing list for discussion of
> asynchronous DBI
> or should discussion continue here on dbi-dev? (now I'm talking crazy)
Here. But...
> I am also proposing to write at least one wrapper driver that
> will provide the functionality against a synchronous DBD.
best to start talking again once you've got something working.
Drivers are, as always, free to add support for anything they like
using the various driver-specific mechanisms provided by the DBI.
For now I'd like to shelve this thread. Thanks to all who contributed.
I'll return to it once I get to address the issue for DBI v2, or
a driver author brings it up because they're actually working on it
(or have a patch contributed by someone who has).
Tim.
> On Thu, 9 Sep 2004 17:52:44 +0100 (BST), Matt Sergeant
> <[EMAIL PROTECTED]> wrote:
> > On Thu, 9 Sep 2004, Dominic Mitchell wrote:
> >
> > > Even if your DB library doesn't provide access to the file descriptor,
> > > you can still use the self pipe trick to work around that.
> > >
> > > http://cr.yp.to/docs/selfpipe.html
>
> --
> David L Nicol
> IT Consulting since 1986
