Re: [Planner Dev] Fixes for gantt bugz # 148637, 149359 and 149651

["lincoln phipps openmutual net" <lincoln phipps openmutual net>]

Richard Hult wrote:
> On tis, 2004-08-10 at 01:33 +0100, lincoln phipps openmutual net wrote:
>
> > Richard,
> > 	
> > 	I've redone  this ...and in the best open source
> > tradition its completely refactored and has new features ;)
>
>
> ;)
>
>
> > I've still got the 10,000 day task limit in as I  think this
> > is an important sanity check which is already in place in
> > the Task dialog.
>
>
> But why? If I'm not mistaken that does't even cover our 2038 limit. I
> think it's rather odd to have two limits, one absolute, that is real,
> and one relative that's just virtual.

OK - I'll remove it ;) I'll patch where its also in the task dialog
too.

> > For my new feature, you can now SHIFT+click drag a Fixed
> > duration task in gantt to update its duration (as opposed
> > to the work) (I always wanted to play with the modifier keys ;)
>
>
> I wonder if dragging a fixed duration task should change the duration
> instead, it will look very odd if dragging a task won't change it's
> visual length.
Yup - OK I'll swap the logic....

Fixed work tasks drag = change in work,
Fixed work tasks SHIFT+drag = change in work (no difference from above),

Fixed Duration tasks drag = change in duration,
Fixed duration tasks SHIFT+drag = change in work (gantt stays fixed
	but work cell changes).

> > That funny safe time value was 1st Jan 2038 so I now use,
> > 	mrp_time_compose( 2038,1,1,0,0,0);
> >
> > as thats more obvious.
>
>
> Hm, that's a pretty expensive function so a hardcoded value with a
> comment is fine :)

Its only used a few times so not too bad. Where can I store
global constants i.e. not defines but const which all programs
can read ? Then I can calculate once and reuse.

> > The reason for both setting the time values to the correct
> > limits outside the loop and then checking for verflow within
> > the loop is that,
> >
> > 	t = planner_scale_time_next (t, priv->major_unit);
> >
> > can actually overflow within the loop.  My checks seemed
> > safer on my testing.
>
>
> Hm, ok.
>
>
> > I've fixed my comments.
> >
> > I've moved this property undo to planner-task-cmd.c/planner-task-cmd.h
> > and it now handles both work and duration updates.
>
>
> Great!
>
> /Richard