Larry, the problem with triggers is performance. If you fire a trigger that logs every time someone accesses a record, you are generating an awful lot of traffic!
Despite the performance hit, the trigger will not work as designed. Any multi row form will not notify the database that row 1234 was accessed until the highlight in the form moves to that specific row of data. Consider: EDIT USING PatientList WHERE LastName = 'Berry' List of patients: Berry, Albert Berry, Chuck Berry, Halle Berry, John Berry, Laurence Berry, Marc Berry, Tanya In a regioned form, the user would be able to read all the information for Chuck, Halle, John, Laurence, Marc and Tanya while only the Albert record would display the hit. Lawrence Lustig <[EMAIL PROTECTED]> wrote: >> 1) �Patients will always have a right to see who has >> accessed their file. � >> That means that the R:Base/R:Tango application >> should keep a log of >> every time a user views confidential information. >> (right before the EDIT >> USING or BROWSE command, e.g., have an INSERT >> command into >> the log table.) �And there should be a way to report >> all access of a >> given individual's records. > >This seems like an enormous amount of work to retrofit >into existing systems, and even more work to make sure >that _every_ time someone accesses data a record is >made. �Also, you can't allow anyone to have access to >the R> prompt since BROWSE, etc commands could be >issued that would result in data being shown. > >What would really work well is a SELECT trigger, so >that everytime a row was accessed you could fire some >code. �You would need some way to find out what >columns are in the output set, because some SELECTs >would be "innocent" but some would need to be logged >(for instance, a group by ZipCode with no identifying >information would not need to be logged, a group by >PatientID would be). > >All you'd need to do is implement a set of robust >SELECT triggers for those tables containing >identifying information. > >I don't know of any database that actually implement >SELECT triggers (although the concept sounds vaguely >familiar, so maybe someone does have it) but any >database that did would be particularly well suited to >HIPAA-subject applications. > >> I'm also planning to nag RBTI a little on a >> long-standing request for a >> new R:Base command: >> >> UNLOAD DATA FOR tableView AS FORMATTED USING column >> x y, >> column2 x y, etc. � >> >> It would be the outgoing counterpart to the existing >> LOAD ... AS >> FORMATTED command. > >Although it would be nicely symetrical to have an >UNLOAD matching the LOAD, am I mistaken in thinking >you can do this in a single command already by >concatenating a bunch of LJS(CTXT(ColName)) >expressions in a SELECT statement? >-- >Larry > >__________________________________________________ >Do you Yahoo!? >Yahoo! Mail Plus - Powerful. Affordable. Sign up now. >http://mailplus.yahoo.com > > -- Albert Berry Full Time Consultant to PSD Solutions 350 West Hubbard, Suite 210 Chicago, IL 60610 312-828-9253 Ext. 32 __________________________________________________________________ The NEW Netscape 7.0 browser is now available. Upgrade now! http://channels.netscape.com/ns/browsers/download.jsp Get your own FREE, personal Netscape Mail account today at http://webmail.netscape.com/
