Hm. Say for example, we are in PST with a 3 hour difference and our test data has HST timestamps from 8:00PM to 10:00PM for 2006-08-23. With the timezone change the dates will change to 11:00PM on 2006-08-23 to 1:00AM on 2006-03-24. Say we are doing a test on DailyProjectData for 2006-08-23, which should access only one day to get the data (although I think it doesn't actually check the value of the date.).
Would solution (2) still work? Logically it shouldn't work. It sort of depends how the data is processed in the Analysis code. thanks, Aaron ----- Original Message ----- From: Philip Johnson <[EMAIL PROTECTED]> Date: Thursday, August 24, 2006 2:10 pm Subject: Re: [HACKYSTAT-DEV-L] TimeZoneChanger To: HACKYSTAT-DEV-L@HAWAII.EDU > --On Thursday, August 24, 2006 1:38 PM -1000 "(Cedric) Qin ZHANG" > <[EMAIL PROTECTED]> > wrote: > > > It's not the test cases only. All analysis code that accesses non- > tstamp UTC values > > needs to be wrapped. At least this is the case with > hackyApp_BuildAnalysis code. But, > > still, I think option (2) involves much less effort than option (1). > > I believe it is the test cases only. Remember that this whole > issue only exists because > we are shipping sensor data collected and stored in one time zone > to a server installed > in another time zone. "Regular" sensor data files do not get > "shipped" in this way, so > the problem does not exist. > > Cheers, > Philip >
