Re: [fluid-dev] Fluidsynth changes

[Josh Green]

On Sun, 2007-04-22 at 17:47 +0200, Miguel Lobo wrote:

> 
> Definitely.  As I mentioned, that name was just meant as a joke and I
> hope nobody took it in any other way.  In fact, I have already renamed
> the fork to something else completely Fluidsynth-unrelated. 
> 

Ok, I hadn't realized that name was a joke ;)


>         Sounds like a good plan.
> 
> Well, I'm not sure now.  It seems you have a clear vision for
> Fluidsynth now and that vision seems incompatible with my own.  I will
> made of course my code public when (and if) it's ready, but I doubt
> Fluidsynth is going to be much influenced by that. 
> 

I think it really depends on if it seems like a good move to go to C++
or not.  If the API that you create seems really nice, there aren't a
lot of added dependencies, still friendly to embedded systems and there
is a way to support C based applications, I'm all for it.

> I know glib.  In fact I programmed a little bit with GObject some
> years ago.  Suffice to say, I didn't like it. 
> 

It definitely took me a while to get used to it.  Back when I forked
Swami to what the development version is currently I had to make a
decision over whether I would use C++ or GObject.  I also made a choice
of which toolkit and researched the ones at the time.  My general
feeling then was that gcc was rather top heavy in regards to C++ (long
compile times, etc).  That's very different today though, and so
sometimes I do question if my choice was a good one or not.
libInstPatch and Swami are completely GObject based, and in my opinion
it is well coded when looked at from an OO perspective.  There is a bit
more boiler plate stuff of course, but when it comes down to it, it
often feels like more important to what language you're using, is good
software design.  I imagine you can shoot yourself in the foot with C++
just as well as you can with GObject.  Having said that, I've often
thought about what it would take to convert libInstPatch and Swami to C
++.  My thinking currently is that more important is making the current
code base complete enough so its a fully functional program, then
perhaps convert it to C++ later, if it seems to be beneficial.  I think
that decision has mostly to do with maintainability of the code though.

> Just curious: are you planning to GObject-ify Fluidsynth?  It seems to
> me that would be as much work as the C++ translation.
> 

The thought has crossed my mind.  Its already wrapped using GObject in
libswami.  I think its not something that is really necessary though.
The FluidSynth API is rather simple in my opinion as it is, so the
question for me is still whether it would really benefit from being
OO-ified.

> Kind regards,
> Miguel
> 

Best regards,
        Josh