Changed Entry Dates after updating from 2.6.4-2 to 2.6.6 and/or 2.6.7

[Wm]

Sat, 11 Jul 2015 20:16:38 
<790D4F52-F1BF-404E-BBF5-ECECEC3A3A77 at ceridwen.us>  John Ralls 
<jralls at ceridwen.us>

>Wm,
>
>Fred’s right that the blank transaction date isn’t from the file. 
>It’s either the date when the register was opened or the last posted 
>date used if a transaction has been created since the register was 
>opened. Unless you think that Detlef is punking us, we can assume that 
>it represents how GnuCash is interpreting the current date — shown in 
>the taskbar in the lower-right of the screen shots — with a 
>1432015200 seconds offset.
>
>Interestingly, log2(1432015200) ~= 30.4, rather close to the size of an 
>int32_t, though both GDateTime, GTimezone, and GnuCash are quite 
>scrupulous about using int64_t for time representations. IIRC so is 
>msvcrt.
>
>It gets more interesting: 2.6.4 used glib-2.28 and 2.6.5 and later use 
>glib-2.32. A significant change between the two is that the newer 
>version has a rewritten Win32 section for gtimezone.c that among other 
>things gets timezone information from the registry. It’s possible 
>that the Win8.1 registry has a different timezone format, but if that 
>were the case I think we would have
>heard about it sooner. The traffic on this list and in Bugzilla 
>indicates that there are lots of users on 8.1. The other possibility is 
>that Detlef’s Registry TZ database has gotten corrupted somehow, 
>though if that were true I’d think the system clock would reflect the 
>problem too and it clearly doesn’t.
>
>I don’t have all of the pieces assembled in my head yet, but I think 
>that’s a more promising line of reasoning that is file format because 
>the blank transaction is affected.

More thinking aloud.

John Morris, this list 2015-06-22
===
Subject: Start Day of Weekly Report
===
had a problem getting his gnc to understand his system time settings.

Possible connection?

-- 
Wm...