Re: Possible timing issue with IO

Hi James

Thanks for the thought. I've considered a batch job, but this has to be done in a synchronous manner, as another process relies on the results being there right away. The other process, from a data warehouse pull, actually, writes out reports based on the fresh data.

Vern

On 1/24/2014 4:44 PM, James H. H. Lampert wrote:

Another approach:

I was asked to write a program that would issue email messages in
response to activity completion in our CRM product (Wintouch). And if
the activity had attachments, it had to include the attachments in the
message.

The only problem is that the exit point from which I would be launching
this was called AFTER the activity record was written, but BEFORE the
attachments had been saved to the IFS.

So after a bit of hemming and hawing, I came up with a solution: the
attachment program checks to see if the "attachment" flag on the record
is set, and if it is, it spawns itself off as a delayed batch job, and
if the spawned-off emailer still didn't see the attachment, it would
sleep (a second at a time, as I recall), while waiting for the
attachment to be written.

Surely, if all else fails (and assuming your duplicate-processing
program deals with the duplicates, rather than just checking for them),
you could spawn it off as a batch job.

--
JHHL