I don't know of the implementation or reasoning here, but there is a lot less memory used if its not instantiating new objects each time. I think we should stick with that approach. We should either say that class level variables are not allowed (like on controllers and servlets) or we say they are allowed but all variables get cleared before a rule is used.
Tammy, what was the fix in 1.5 for this? Is there a ticket? Was there a regression test for it originally? Ben On Thu, Oct 13, 2011 at 5:46 PM, Tammy Dugan <[email protected]> wrote: > It also corrupted 3 days worth of patient data across 4 clinics so > needless to say we were pretty unhappy that a problem we fixed in 1.5 came > back in 1.7 and corrupted patient data. > > Tammy > > > On 10/13/2011 11:27 AM, McKee, Steven Jay wrote: > > Darius,**** > > ** ** > > CHICA encountered a major issue with Logic yesterday. We had information > get crossed between multiple patients, and we tracked it down to a thread > safety issue in Logic. Currently the Token Service’s “getRule(String)” > method (in OpenMRS 1.7) retrieves a rule from a rule provider and then > caches it. The next time the rule is asked for, it gets pulled directly > from the cache. Our Java rules translated by the Arden translator had class > level variables that were getting changed by concurrent threads, and we were > seeing incorrect data for patients.**** > > ** ** > > In OpenMRS 1.5 the logic service would cache the rule, but it would return > a new instantiation of the rule class so it was thread safe. Now in 1.7 + > the thread safety no longer exists because the same instance of the rule is > directly returned from the cache. **** > > ** ** > > This is a very high priority issue that needs to be resolved immediately.* > *** > > ** ** > > Thanks,**** > > ** ** > > Steve McKee**** > > Children’s Health Services Research**** > > Indiana University School of Medicine**** > > Phone: 317-278-9660**** > > Email: [email protected]**** > > ** ** > ------------------------------ > Click here to > unsubscribe<[email protected]?body=SIGNOFF%20openmrs-devel-l>from > OpenMRS Developers' mailing list > > > -- > Tammy Dugan > CHIRDL Technical Lead > Children's Health Services Research > IU School of Medicine > > ------------------------------ > Click here to > unsubscribe<[email protected]?body=SIGNOFF%20openmrs-devel-l>from > OpenMRS Developers' mailing list > _________________________________________ To unsubscribe from OpenMRS Developers' mailing list, send an e-mail to [email protected] with "SIGNOFF openmrs-devel-l" in the body (not the subject) of your e-mail. [mailto:[email protected]?body=SIGNOFF%20openmrs-devel-l]
