Re: [AD] Audio plugin function names

[Evert Glebbeek <eglebbk@xxxxxxxxxx>]

On 27 Sep 2008, at 20:00, Peter Wang wrote:
> > Right now in the audio plugin, functions are named
> > al_sample_play, al_sample_stop, al_voice_destroy,
> > al_sample_data_destroy etc.
> > Shouldn't those be called al_play_sample, al_stop_sample,
> > al_destroy_voice, al_destroy_sample_data etc. to be consistent with
> > the rest of the API?
>
> Yup.

If no one else has started doing this yet I can look into it during  
the week. I'll be moving as well, but hopefully I'll still have some  
time.
So should that be al_play_sample or al_audio_play_sample? If no one  
expresses a clear preference I'll go with al_play_sample.

>   Also need to get rid of, or hide away, most of the "audio
> property" functions and replace them with specific functions for each
> property.  It's very hard to tell which properties apply to which  
> audio
> objects and which are read-only.

You mean the al_sample_get_float/long/pointer/enum etc. functions? I  
have to admit I didn't quite get those from a cursory look at the  
documentation.

That also brings me to a slight complaint that I have about the  
current audio system: it's a bit hard to do simple things.
Most of the time I just want to load a sample from disk and play it  
when needed. To set things up to do that in Allegro 4 I would

- initialise audio subsystem
- load sample
- play sample

now judging from the examples it seems I have to

- initialise audio subsystem
- create voice
- create mixer
- attach mixer to voice
- load sample data
- create sample from sample data
- attach mixer to sample
- play sample

Steps two and three involve passing parameters that I have no real  
idea off what they do (I know, read the documentation) and I would be  
happy to go with a simple "pick whatever you think best, I don't care  
as long as it works".

I'm not saying that this greater level of control is bad (far from  
it) and obviously people do want it. But surely there has to be a way  
to make the simple case easier without compromising the more advanced  
features? For instance, a function that could combine steps 2, 3 and  
4 and (more importantly) one that could combine 5 and 6?
I don't know how that would work with freeing the different objects  
that get to point to eachother in the above chain.

Evert