[Kde-bindings] cleaning up smoke

[Richard Dale]

On Friday 02 March 2007, Caleb Tennis wrote:
> Does anyone object to my C++ifying smoke up a little bit more nicely?
>
> A few things that come to mind are:
>
> * Using function bindings instead of function pointers
> * Migrate from using pointers and indexes to managed lists (std::vectors,
> most likely) * A few other miscellaneous odds and ends
Well I'm not sure if 'C++ifying smoke' should be and end in itself, as the
bindings languages we interface with like Ruby or C# have C apis for that. My
concern about std::vector is that it will introduce build fail problems we
wouldn't otherwise have. I might be completely wrong and it works great with
every C++ compiler out there including Windows - I just don't know enough
about the subject. What do function bindings do that function pointers don't?

Will this change do one or more of the following:
- Fix existing bugs
- Make smoke easier to use by the handful of people who currently use it
- Add to current functionality
- Make it easier to add functionality in the future
- Speed up smoke

As long achieves at least one of those, and it doesn't mess anything up I'm
fine with it.

But we're just getting qtruby to build with cmake on Mac OS X and Windows and
I would like to do a new release in the next two weeks or so. And also
finally do a binary gem because I'm really getting worn out with everytime
qtruby gets mentioned on ruby-talk people say how difficult to build it is
and suggesting using other gui toolkits instead.

So I'm not very keen on the Smoke lib changing before then. Would it be
possible to do the changes in a branch, or in a patch so we can put it off
until after the qtruby release?

-- Richard