path: root/src/corelib/tools/qtimezoneprivate_win.cpp
diff options
authorEdward Welbourne <>2018-01-18 13:48:26 +0100
committerEdward Welbourne <>2018-02-07 12:21:00 +0000
commit6d3e5ac6d2f34a0da609f863bed95d3be571c589 (patch)
treef372cab569443cf05514cc4a6092cb0b6400e958 /src/corelib/tools/qtimezoneprivate_win.cpp
parentaec6fae60b4e229b7e00e534c47d341caf49cddc (diff)
Prefer ICU over system info on MS-Win
The ICU data is more complete; and its APIs allow it to tell the whole story, which MS-Win's APIs can't (e.g. Morocco's otherwise-normal DST has a hole in it for Ramadan; when this makes four transitions in a year, the MS-APIs can't report more than two, so ignore the Ramadan gap); and their design obliges us to use heuristics to kludge round mis-description of simple changes to standard time in non-DST zones, which can't be distinguished from certain (more) obscure cases of changes to DST coinciding with a change to standard offset (causing a DST transition to hide, leaving its other end *looking like* a non-DST change to standard offset). Using ICU, when available, reduces how many end users shall be mislead by such kludges. [ChangeLog][Windows][QTimeZone] Changed MS-Win to use ICU for time-zone data, when available, in preference to MS's TZ APIs. The choice is made when building Qt. This will give reliable results when non-DST transitions arise, or when a zone's DST is not simple (e.g. interrupted by Ramadan). Task-number: QTBUG-42021 Change-Id: I9cdd65713ecdaf8cce52dd924fbc7371630977c8 Reviewed-by: Thiago Macieira <>
Diffstat (limited to 'src/corelib/tools/qtimezoneprivate_win.cpp')
1 files changed, 1 insertions, 0 deletions
diff --git a/src/corelib/tools/qtimezoneprivate_win.cpp b/src/corelib/tools/qtimezoneprivate_win.cpp
index 8bde07c710..ec91b7e8a8 100644
--- a/src/corelib/tools/qtimezoneprivate_win.cpp
+++ b/src/corelib/tools/qtimezoneprivate_win.cpp
@@ -49,6 +49,7 @@
#ifndef Q_OS_WINRT
+// The registry-based timezone backend is not available on WinRT, which falls back to equivalent APIs.