[vlc-devel] [RFC] Regarding the use of *_Control()
jpsaman at videolan.org
Tue Apr 15 11:35:16 CEST 2008
Pierre d'Herbemont wrote:
> On Apr 15, 2008, at 8:57 AM, Jean-Paul Saman wrote:
>> Indeed my answer is somewhat subjective.
> Thanks :)
>> What you propose is to break
>> module ABI. I got the impression that we as team didn't want that.
>> proposal only wraps the *_Contrlol() function without really
>> the module ABI.
> True. Yet, first step would be just to have a nicer accessor. Second
> step is to break the ABI by hiding _Control()
>> If we are going down the road of redifining (and breaking the module
>> ABI), then I propose we do it proper by rethinking the way the
>> *_Control() functions are defined. The goal is then to end up with a
>> well-defined new module ABI (which is better maintainable and harder
>> misues). IMHO that would be better, then just wrapping the old
>> *_Control() and thus adding a new layer of indirection.
> I'll think about it. This one is a bit harder, as more code needs to
> be changed in an non automatic way. But for sure, we could rethink it.
Well then we can think about fixing pause/frame accurate/last seconds
not being played errors (and all that are related to current structure).
> Do we agree that I should remap *_Control() as a first step?
I think it is premature to do that now. It is my feeling that there are
still things (fundamentally) very wrong or broken in git. Let us focus
on fixing that first, before rewrite internal module ABI's. Which will
definitely destabilize vlc more (at least temporarily) and make it
harder to get git in a sensible state.
> Eventually an other RFC will follow.
First RFC then remapping and rewritting (has my preference).
> Thanks for your time,
No problem. Just my 2cts.
More information about the vlc-devel