uPortal IRC Logs-2012-10-17

Owned by Apereo Infrastructure
Last updated: Oct 17, 2012
2 min read

[10:16:11 CDT(-0500)] <EricDalquist> this event data stuff is really not fun

[10:16:21 CDT(-0500)] <EricDalquist> I think I can do a drop-in fix

[10:16:26 CDT(-0500)] <EricDalquist> that won't break things for anyone

[10:16:50 CDT(-0500)] <EricDalquist> but will result in a 1 day jump in stats data for people on a negative tz offset

[10:17:14 CDT(-0500)] <athena> ahh

[10:17:20 CDT(-0500)] <athena> well, that'll probably be ok

[10:17:23 CDT(-0500)] <EricDalquist> yeah

[10:17:33 CDT(-0500)] <athena> whole thing is frustrating though

[10:17:34 CDT(-0500)] <EricDalquist> I'll have to see how hard it woudl be to shift the existing events forward one day

[10:17:35 CDT(-0500)] <EricDalquist> yes

[10:17:44 CDT(-0500)] <EricDalquist> well and the fix isn't perfect

[10:17:52 CDT(-0500)] <EricDalquist> it uses the local TZ for storing LocalDate objects

[10:17:56 CDT(-0500)] <EricDalquist> and UTC for everything else

[10:18:00 CDT(-0500)] <EricDalquist> which seems to work

[10:18:15 CDT(-0500)] <EricDalquist> since the LocalDate is the only one that is missing the hours

[10:18:19 CDT(-0500)] <EricDalquist> which is the cause of the problem

[10:27:47 CDT(-0500)] <athena> well, if it works (smile)

[10:31:57 CDT(-0500)] <EricDalquist> yeah

[10:32:02 CDT(-0500)] <EricDalquist> that is my goal at this point ...

[10:57:46 CDT(-0500)] <EricDalquist> so I think my plan right now is to fix the LocalDate mapping

[10:58:03 CDT(-0500)] <EricDalquist> the full timestamps in the DB will remain as UTC

[10:58:36 CDT(-0500)] <EricDalquist> and it should just result in a hole in the stats aggregation equals to the UTC offset of the portal jvm

[11:02:44 CDT(-0500)] <athena> sounds reasonable

[11:04:46 CDT(-0500)] <athena> do you have any idea what happens if a DLM fragment has more than one evaluator defined?

[11:37:46 CDT(-0500)] <EricDalquist> I'm not sure

[13:27:34 CDT(-0500)] <EricDalquist> ok .... I think this fix will work

[13:27:42 CDT(-0500)] <EricDalquist> getting a vm setup to test the various TZ offsets :/

[13:31:03 CDT(-0500)] <athena> aw (sad)

[14:44:07 CDT(-0500)] <EricDalquist> woot

[14:44:08 CDT(-0500)] <EricDalquist> ok

[14:44:14 CDT(-0500)] <EricDalquist> so tested the jvm tz fix

[14:44:19 CDT(-0500)] <EricDalquist> if you have a UTC offset >= 0

[14:44:28 CDT(-0500)] <EricDalquist> the net effect is zero

[14:44:36 CDT(-0500)] <EricDalquist> this bug only affects people with a UTC offset < 0

[14:45:00 CDT(-0500)] <EricDalquist> and we will just have to live with our old aggr data shifted back in time by 1 day

[14:45:06 CDT(-0500)] <EricDalquist> and a gap in our stats