[vlc-devel] Re: vlc: svn commit r20432 (pdherbemont)

[Pierre d'Herbemont]

On 7 juin 07, at 15:06, Damien Fouilleul wrote:

> Subversion daemon wrote:
>> r20432 | pdherbemont | 2007-06-07 14:34:50 +0200 (Thu, 07 Jun  
>> 2007) | 2 lines
>> Changed paths:
>>    M /trunk/src/control/core.c
>>    M /trunk/src/control/event.c
>>    M /trunk/src/control/input.c
>>    M /trunk/src/control/libvlc_internal.h
>>
>> Libvlc Event: Add support for input event.
>>
>>  > http://trac.videolan.org/vlc/changeset/20432
>>
>>
> i'm not sure it is a good idea to send an event for input  
> 'position', as this has the potential of generating a lot of events  
> and stall the input thread. Just like the GUI interfaces currently  
> do, 'position' should be polled rather then pushed.
> i'd rather see input 'state' changes, playlist 'item' changes,  
> etc... as events.

For now, position event is a 'fixme'... But I do understand your point.

I have in mind two things:
- Felix pointed that we might want to be able to set a custom  
position resolution. So that we don't have the position event  
triggered every 40ms when only 1/2 s is needed.
- About stalling the input thread, we can simply detach a new thread  
that will call the client callbacks. I'll try to work on it.

About polling I am not convinced that it is a good solution. Most of  
the time you end up in wasting CPU time. (especially when polling for  
an event that never occurs). Thus I really think it is fair to  
consider a position event.

> do not misunderstand me, i do appreciate the work, and i will  
> change the activex/mozilla controls shortly to support libvlc events.

No prob... Nice to ear some feedback :)

Pierre.

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html