[vtk-developers] Updates to VTK's coding standard page

[David Gobbi]

On Thu, Mar 14, 2013 at 3:11 PM, Will Schroeder
<will.schroeder at kitware.com> wrote:
> Kyle to a large extent it's historical. Twenty years ago many things were
> brittle including compilers, STL, and wrapping. This led to a desire to keep
> things clean and simple. For example the wrapping process choked for a long
> time on fancy STL stuff. I also am of the opinion that the API to STL is,
> shall we say, poor plus it affects compilation time. We wanted to hide some
> of the nastiness from the wrapping process, from users, and encourage
> passing data through simpler exchange mechanisms. The PIMPL approach also
> hides the underlying implementation which provides more freedom if you want
> to change things up at a later time. I'm sure I'm forgetting a couple of
> other reasons, but Bill is in Santa Cruz for the month so I doubt he'll chip
> in ;-)

You say twenty years ago?  Even ten years ago all the most widely-used
compilers were fairly brittle.  Microsoft is very wishy-washy its standards
support up to this very day.

Anyway, my main issue with STL is that, currently, APIs that use STL
are not wrapped.  And in my mind, that deficit strongly outweighs any
benefit.  Of course we could wrap a chunk of the STL, but I don't know
where the resources for such a project would come from.  In any case
I'd rather be conservative about the interface just to keep it consistent.

 - David