Re: [lmi] Anomaly: empty messagebox

[Greg Chicares]

[BTW, the anomaly reported earlier was reproducible with an
lmi release from last August, and with cygwin as well as wine.]

On 2017-01-28 12:28, Greg Chicares wrote:
[...]
> Could it be that this is InputSequenceEntry::DoOpenEditor()
> showing nothing at all? IOW, is this the intended behavior of
> the following code path?
> 
>     if(intervals.empty())
>         {
>         // have single row (initial state)
>         add_row();
>         return;
>         }
> 
> If that's the explanation, then maybe the right thing to do
> is to establish !intervals.empty() as a ctor postcondition;
> I was thinking of doing that anyway.

Indeed, if I do that, the empty-messagebox anomaly seems no
longer to be reproducible, and then the block of code quoted
above could be removed or replaced with an assertion.

The postcondition does seem reasonable--and, in fact, its
absence seems unreasonable--because specifying nothing can
only mean that a default value is to be used from inception
through maturity. Even if no value is specified, those
endpoints still exist, so the interval between them exists,
and therefore at least one interval must always exist.