Chuck,

On 25. 2. 2015, at 17:39, Chuck Hill <[email protected]> wrote:

> That should not be related.  The only two causes of this that I have see are:
>       • Implementing awakeFromFetch() and triggering a fetch in that method

Hardly: the only awakeOfFetch of mine just flushes some internal caches.

>       • An exception that leaves EOF in a dirty state with the fetch still 
> marked as in progress

Hmmm... this just _might_ happen, for I did get one NPE before; duly catched 
and reported -- it is in place where imported CSV is processed and exceptions 
happen sometimes, e.g., when the CSV is malformed. But it is possible that this 
time the reason was not a malformed CSV, but some mess in EOs instead.

Alas, since exceptions are comparatively common in there, I reported them 
without the full stack, so I know just the CSV line on which it happened, which 
does not help much :) Meantime of course I've uploaded a new release which does 
print full stack -- no need to add that from that time the problem never 
reappeared :/

But it looks like a comparatively plausible possibility, thanks!

> My advice is to catch the notification and log a stack trace as a first step 
> to tracking this down.

Might add that one, too. Thanks again!

All the best,
OC

> On 2015-02-25, 1:40 AM, "OC" wrote:
> 
> Chuck,
> 
> On 23. 2. 2015, at 20:01, Chuck Hill <[email protected]> wrote:
> I’d use a different EOF stack but NOT from 
> ERXObjectStoreCoordinatorPool.maxCoordinators which gets shared between 
> requests.  You want one that is not shared and is dedicated to the import 
> task.  
> For this, just
> EOObjectStoreCoordinator osc = new EOObjectStoreCoordinator();
> ERXEC ec = ERXEC.new(osc);  // er something like that just typed this from 
> memory
> 
> Hmmm... I did, and my testing went well. From the server though now I'm now 
> getting reports
> 
> ===
> A fault was fired while another fetch was in progress and no other channels 
> were available to process the fault.  Either make sure not to touch faults 
> while fetching, or register more channels with the EODatabaseContext (perhaps 
> in response to the EODatabaseChannelNeededNotification). Original 
> Exception:N/A
> ===
> 
> Does that make any sense? Indeed I am not observing 
> EODatabaseChannelNeededNotification and creating channels on-demand; on the 
> other hand, till now, I have never seen this problem, not even when I used to 
> set ERXObjectStoreCoordinatorPool.maxCoordinators>1.
> 
> Might it be related?
> 
> Thanks,
> OC
> 
> 


 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list      ([email protected])
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com

This email sent to [email protected]

Reply via email to