Closed nikhilgupta10 closed 8 years ago
nikhilgupta10 imported these comments from Sourceforge: The user dchassin does not exist anymore. Therefore assigning this to afisher1. "dchassin": * status changed from new to accepted
Confirmed. If TZ is not set timestamp_set_tz tries to guess the default timezone by looking at _tzname[]. It now appears that instead of returning PST8PDT like it used to, it now returns \Pacific Standard Time8Pacific Daylight Time. This is new behavior and must be addressed in the guess code.
No known \easy\ fix yet.
Workaround: suggest setting TZ for now to avoid guessing behavior.
Fix option 1: treat these as bad specs and default to UTC0 when TZ is not set and long form zones are encountered. This may result in broken models until TZ is universally set.
Fix option 2: add an index to convert long names to short names in tzinfo.txt. This would require researching what MS uses for long names around the world.
,
"dchassin": * status changed from accepted to closed
The problem is that ftime() changes _tzname when it is called on Windows system. From tests done it appears several time function do this. The behavior undocumented and was reported to Microsoft. The current fix is to copy the values of _tzname before any time function is called. Changeset:3422 implements this fix.
,
Under trunk, GridLAB-D ignores all timezone specifications. What's more, it tries to \guess\ with an invalid specification that isn't listed in the tzinfo.txt. All files tested appeared to do this, but test files are really needed, use the ones of #595.
Untested under Linux/Mac, so maybe it works there.
,