I have many of the same questions you do about Workers. I only discovered the
new concept of Workers three days ago and am running experiments to try to
figure them out myself. But I can offer up the following.
As for High Velocity: I created an iteration loop that called a Worker. The
only thing that I had the Worker do was to Create a new record, store the
iteration number + current date + current time in it and save the record. I
set it up for 500,000 iterations. It took about the same time as just running
in the same process (I didn’t exhaustively compare numbers). It doesn’t seem
to have the normal time lag that would occur to start up a New Process.
Anyway, it was very fast.
I was also concerned about what would happen when a Worker was given a new
request while it was still executing a previous request. So I lmited the
integration to just 20 records, but introduced a 3 second time delay. It
stored up those requests and executed all of them. It did not drop any of the
request on the floor. I believe it executed the Worker requests in the same
sequence of the Worker calls from my iteration loop, but I am not certain about
that. I now am wondering what the limit is for the number of requests to the
Worker that can be stored up?
I did discover what I think is a Worker bug in doing these simple tests. Each
time I ran these little tests, the very last record that was created and then
saved by the Worked is edit locked. It can’t be deleted.
I hope this helps at least a little bit.
David Belote
Colorado Springs
On 10/16/16, 10:18 AM, "4D_Tech on behalf of Foucauld Perotin"
<4d_tech-boun...@lists.4d.com on behalf of fp.lis...@teluric.com> wrote:
Hi,
I have some questions about workers. Sorry if it is obvious for some of
you. I still don’t feel so comfortable with this new stuff.
- May the workers be local processes? (naming them with a $, I guess)
- What about the stack size ? I see nothing about that, and I do not
understand why. I guess the stack will have the mystery "default size", but I’m
not sure... ;)
- I know that preemptive execution implies no interprocess variables. OK.
But, if the preemptive aspect is not a priority for me now, can I still use
inteprocess variables without any problem in my workers?
– Is the use of workers not for a high velocity compute section of an app,
but for a regular part, with basic records filling and so, reasonable or not?
with the idea to call the worker-process by the "call form" command, when
needed). Why would I do that? Because some of my apps have a concept of several
permanent processes with reuse of those. So, the workers would be better than
the processes which are most of the time sleeping, then wake up, then go back
to sleep, and so on...
Thanks for your upcoming replies! :)
--
Foucauld Pérotin
Try again.
Fail again. Fail better.
Samuel Beckett
**********************************************************************
4D Internet Users Group (4D iNUG)
FAQ: http://lists.4d.com/faqnug.html
Archive: http://lists.4d.com/archives.html
Options: http://lists.4d.com/mailman/options/4d_tech
Unsub: mailto:4d_tech-unsubscr...@lists.4d.com
**********************************************************************
**********************************************************************
4D Internet Users Group (4D iNUG)
FAQ: http://lists.4d.com/faqnug.html
Archive: http://lists.4d.com/archives.html
Options: http://lists.4d.com/mailman/options/4d_tech
Unsub: mailto:4d_tech-unsubscr...@lists.4d.com
**********************************************************************
