On Mon, Apr 09, 2001 at 03:31:19PM +0200, Juergen Vigna wrote:
> Well I couldn't as IMO I'm more illiterate on STL (in general) the you :)
>
> But there are general problems in this type of stuff. F.ex. you can have a
> look at the AppendRow calls in tabular.C. You see that I do explicitely
> insertion of a row instead of using the (#ifdefed out) version. Why?
> That's because if I use the standard STL .insert it won't work (it should
> but it doesn't!). So maybe we have here a similar problem. The problem
> always regards the insertion of some stuff or the assignment of some stuff.
>
> The problem in AppenRow I had (so I remember vaguely) is that instead of
> only inserting a new item it reinitialized all the vector and lost so all
> the informations inside the InsetText's :(
I had no problem when I enabled the STL code in AppendRow.
Do you have a problem when you compile without optimization ?
Do we want to support buggy compilers ?
BTW you can use
cell_info.insert(cit, vector<cellstruct>(columns_,));
instead of
cell_info.insert(cit, vector<cellstruct>(columns_, cellstruct()));
)