[Kde-pim] KCal::Calendar sortEvents

[David Jarvie]

On Sat 4 October 2008 14:31:47 Allen Winter wrote:
> On Friday 03 October 2008 03:26:59 Ron Goodheart wrote:
> > Well, digging a little deeper...
> > I think the sort code is actually okay, I think KDateTime is at fault.
> > It comes down to a slight difference in how KDateTime and QDateTime
> > handle date equality if one of the dates is a 'date only' type of date.
> >
> > QDateTime considers the 'date only' to be less than a date that has a
> > time. KDateTime does not because it only compares the dates (unless the
> > time zones are different then it changes the time if the first date is
> > 'date only').

QDateTime in Qt 4.4 doesn't have a concept of 'date only', so you can't 
compare it with KDateTime in this respect.

> > The operator less-than code in KDateTime didn't look right so I ran a
> > small test with the following results:
> >
> > dateOnly = "2008-10-02T00:00:00-07:00"
> > datetime =  "2008-10-02T00:01:00-07:00"
> > dateOnly isDateOnly? =  true
> > datetime isDateOnly? =  false
> > dateOnly <  datetime =  false
> > dateOnly <= datetime =  true   <---
> > dateOnly >  datetime =  false
> > dateOnly >= datetime =  true   <---
> > dateOnly == datetime =  false
> > datetime <  dateOnly =  false
> > datetime <= dateOnly =  true  <---
> > datetime >  dateOnly =  false
> > datetime >= dateOnly =  true  <---
> > datetime == dateOnly =  false
> >
> > This is clearly incorrect!
> >
> > With the attached patch (with old code commented) I get:
> > dateOnly = "2008-10-03T00:00:00-07:00"
> > datetime =  "2008-10-03T00:01:00-07:00"
> > dateOnly isDateOnly? =  true
> > datetime isDateOnly? =  false
> > dateOnly <  datetime =  true    <---
> > dateOnly <= datetime =  true  <---
> > dateOnly >  datetime =  false
> > dateOnly >= datetime =  false
> > dateOnly == datetime =  false
> > datetime <  dateOnly =  false
> > datetime <= dateOnly =  false
> > datetime >  dateOnly =  true    <---
> > datetime >= dateOnly =  true   <---
> > datetime == dateOnly =  false
> >
> > I don't understand why the code makes a 'date only' end-of-day unless
> > there was some kind of assumption of start-date and end-date? It does
> > have an interesting side effect, but clearly not desirable.

A date-only KDateTime value represents any time during the day from the 
beginning of the day to the end, so in effect represents the whole day in the 
context which you're discussing ( see the API documentation for 
KDateTime::operator<() ). This should explain the results you see.

> > Are there automated tests for KDateTime that I should be checking?
>
> Ron,
> There is a unit test for KDateTime in kdelibs/kdecore/tests/kdatetimetest.c
>
> Make sure you run cmake with -DKDE4_BUILD_TESTS=ON
> Then you should be able to go in the kdelibs builddir/kdecore/tests
> and run ./kdatetimetest
>
> David Jarvie (who I'm sure is reading this) is the KDateTime guru.

I just got back from holiday this evening and saw this.

-- 
David Jarvie.
KAlarm author and maintainer.
http://www.astrojar.org.uk/kalarm
_______________________________________________
KDE PIM mailing list kde-pim at kde.org
https://mail.kde.org/mailman/listinfo/kde-pim
KDE PIM home page at http://pim.kde.org/