RE: Matlab equivalents needed (yes very original)

[Michael V]

If you run many short simulations then PIDs may get recycled.
But hostname will work with a bit of a tweak. Thanks.

On the other side, looking at mkstemp C code, it is very simple.
Should be doable in Matlab in short order.

Michael


> Date: Mon, 28 Jun 2010 18:23:25 -0700
> From: address@hidden
> Subject: RE: Matlab equivalents needed (yes very original)
> To: address@hidden; address@hidden; address@hidden
>
>
> --- On Mon, 6/28/10, Michael V <address@hidden> wrote:
>
> From: Michael V <address@hidden>
> Subject: RE: Matlab equivalents needed (yes very original)
> To: address@hidden, address@hidden
> Date: Monday, June 28, 2010, 5:26 PM
>
>
>
>
> So are any of these cluster-safe? If I have say 160 CPUs running 160 processes (1 per CPU) and
> all processes write to same directory on same hard drive, is there a chance of conflict?
> Also, is there an implementation of these for Matlab? I do not want to maintain two codebases.
>
> Thanks,
>
> Michael
>
>
> Date: Mon, 28 Jun 2010 17:38:55 -0500
> Subject: Re: FW: Matlab equivalents needed (yes very original)
> From: address@hidden
> To: address@hidden
>
>
> For (2) do any of these solve the issue?
>
> tmpnam
> tmpfile
> mkstemp
>
>
>
> The New Busy think 9 to 5 is a cute idea. Combine multiple calendars with Hotmail. Get busy.
>
> -----Inline Attachment Follows-----
>
>
> Something like <hostname>.<PID> is sufficient for a unique ID - because hosts have unique name in a cluster, and PID is unique in a host.
>
> Maybe <hostname>.<PID>.<current_time_since_the_epoch> .
>
> Regards,
> Sergei.
>
>
>

---

The New Busy is not the too busy. Combine all your e-mail accounts with Hotmail. Get busy.