Rodrigo,
I think that by
In fact most of transaction isolation theory is concerned
with how reads will be affected by writes
David meaned that it avoids collisions, or if you prefer it assures security in atomic data level calls. What could happen if
someone would write the same entities at the same time. They are setting for these kind of things at the data level (DB specific if
IIRW) but this is a pain OFBiz is trying to avoid. I may be wrong but in such cases I'm sure David will correct/complete me...
Jacques
From: "Rodrigo Lima" <[EMAIL PROTECTED]>
Transactions for search?
2008/3/19, David E Jones <[EMAIL PROTECTED]>:
Skip,
Transactions are just as important for reading as they are for
writing. In fact most of transaction isolation theory is concerned
with how reads will be affected by writes.
In OFBiz all major operations are wrapped in a transaction by default,
even rendering of screens for outgoing information (typically in a
separate transaction than the events processing incoming data). Most
entity engine operations get wrapped in a transaction if there is not
one in place, but it can't do that with the EntityListIterator calls
because the transaction needs to stay open until the ELI and the
ResultSet associated with it is closed.
In a busy server environment you'll probably run into database errors
related to this, but it depends on your traffic and the database you
use.
But yes, this IS important and was added because production errors
became common before this was added.
-David
On Mar 18, 2008, at 8:48 PM, [EMAIL PROTECTED] wrote:
> Jacopo
>
> I guess my question should be rephrased.
>
> "Why does the author think that all findListIteratorByCondition()
> calls
> should be surrounded with a transaction, particularly in this case
> where it
> is read-only."
>
> I just want to make sure I am not missing something. I am quite
> happy to
> comment out the exception. In this particular case, 150gigs of log
> file
> were created in one service call to compute stock needs with real
> live data
> from a big company. I am trying to debug this code and need the
> regular
> debug output to be sure things are going as I expect.
>
> If I really need the begin/end, I am happy to do that, I just wanted
> to know
> why.
>
> Skip
>
> -----Original Message-----
> From: Jacopo Cappellato [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, March 18, 2008 7:13 PM
> To: [email protected]
> Subject: Re: findListIteratorByCondition Error
>
>
> [EMAIL PROTECTED] wrote:
>> More on this error. Here is the code in GenericDelegator which
>> writes the
>> error to the log.
>>
>> //throwing an exception is a little harsh for now, just
> display
>> a really big error message since we want to get all of these fixed...
>> Exception newE = new Exception("Stack Trace");
>> Debug.logError(newE, "ERROR: Cannot do a find that
>> returns an
>> EntityListIterator with no transaction in place. Wrap this call in a
>> transaction.", module);
>>
>>
>> As the author desired, it is a REALLY big error. I was writing
>> stdout to
> a
>> file which almost filled up my hard disk with these errors.
>>
>> Could the author comment on why it is important to do this:
>>
>> TransactionUtil.begin();
>> Map inventory =
>> dispatcher.runSync("getInventoryAvailableByFacility",
>> UtilMisc.toMap("productId", productId, "facilityId", facilityId));
>> TransactionUtil.commit();
>>
>> I.E. put the begin/commit around the runSync call other than just
>> to avoid
>> filling up your hard disk?
>>
>
> I'd suggest to avoid redirecting stdout to a file.
>
> Jacopo
>
>> Skip
>>
>>
>>
>>
>> No virus found in this outgoing message.
>> Checked by AVG.
>> Version: 7.5.519 / Virus Database: 269.21.7/1331 - Release Date:
>> 3/16/2008
>> 10:34 AM
>
>
> No virus found in this incoming message.
> Checked by AVG.
> Version: 7.5.519 / Virus Database: 269.21.7/1331 - Release Date:
> 3/16/2008
> 10:34 AM
>
> No virus found in this outgoing message.
> Checked by AVG.
> Version: 7.5.519 / Virus Database: 269.21.7/1331 - Release Date:
> 3/16/2008
> 10:34 AM
>
