Re: [lmi] [PATCH] Use correct type for the first column renderer in the census view.

[Vadim Zeitlin]

On Mon, 31 Aug 2015 22:11:28 +0000 Greg Chicares <address@hidden> wrote:

GC> On 2015-08-28 22:39, Vadim Zeitlin wrote:
GC> > 
GC> >  The trivial attached patch gets rid of many warnings when using lmi with
GC> > wxGTK (or, I assume, with wxOSX, which has similar code, but I haven't
GC> > tested this). The problem was not detected when using the generic version
GC> > of wxDataViewCtrl in wxMSW (but this is going to change with my upcoming
GC> > changes to wxWidgets which will harmonize this code across all ports), but
GC> > in the ports using native control implementation the type of the renderer
GC> > and that of the value returned from the model really must be the same.
GC> 
GC> Okay, it really has been "long" all along:
GC> 
GC>     if(col == Col_CellNum)
GC>         {
GC>         return "long";
GC>         }
GC> 
GC> Committed 20150829T0005Z, revision 6276.

 Thank you! FWIW I've already implemented (but not committed yet as I'd
like to test them a bit more) the changes to wxMSW mentioned above, so if a
similar problem ever happens again, we would notice it even when using this
port. But I also think that I should test lmi under Linux more often, now
that it has at least one potential user under this OS, it's IMHO worth
doing it even more than before -- but please let me know if you disagree.

 Finally, I'd like to say that I still remember about the remaining problem
with the column widths for the premium columns, I just didn't manage to
understand where did it come from, as the code seems correct to me -- but
clearly doesn't work. In fact, I fixed the above problem and other
(autotools build system-related) problems under Linux just because I wanted
to test it there as well, but unfortunately this didn't help me to
understand it any better. I'll continue debugging it tomorrow.

 Regards,
VZ