[vlc-devel] [PATCH v2 01/10] core: add atomic refcounter helper

[Steve Lhomme]

On 2018-07-02 17:44, Rémi Denis-Courmont wrote:
> Assertion in headers, or generally dependencies on config.h in headers

In fact I think we should use precompiled headers and config.h would 
definitely be in there for pretty much (if not all) the files we compile 
in the core/modules. But that's a different subject...

>
> Le 2 juillet 2018 18:34:49 GMT+03:00, Steve Lhomme <robux4 at ycbcr.xyz> 
> a écrit :
>
>     On 2018-07-02 17:33, Rémi Denis-Courmont wrote:
>
>         I have to disagree and your bugs are not an excuse for
>         somebody else to add bugs. Fix your bugs. 
>
>
>     What are you talking about ?
>
>         Le 2 juillet 2018 18:30:31 GMT+03:00, Steve Lhomme
>         <robux4 at ycbcr.xyz> a écrit : On 2018-07-02 17:25, Rémi
>         Denis-Courmont wrote: 1) This creates obvious overflow.
>         Exactly the bug that the inspiring kernel code is meant to
>         fix... 2) So this won't work for e.g. picture_t. Then it
>         won't. 3, 4) Inspiring kernel code does address those issues.
>         And then, no assert in public headers. We already have 3 of
>         these (one from me, two from Thomas). In VLC core headers it's
>         fine as it's a controlled environment, even for external code.
>         For libvlc it wouldn't be OK at all. Le 2 juillet 2018
>         18:18:27 GMT+03:00, "Hugo Beauzée-Luyssen" <hugo at beauzee.fr> a
>         écrit : On Mon, Jul 2, 2018, at 8:12 AM, Rémi Denis-Courmont
>         wrote: Same problems as before. -- Envoyé de mon appareil
>         Android avec Courriel K-9 Mail. Veuillez excuser ma brièveté.
>         ------------------------------------------------------------------------
>         vlc-devel mailing list To unsubscribe or modify your
>         subscription options:
>         https://mailman.videolan.org/listinfo/vlc-devel - type sizes
>         (uint, uintptr...), The serie only affect code that was using
>         uint, except for patch 4 that was using uintptr, and I believe
>         uint to be a better fit; and patch 10. I'm fine with dropping
>         patch 10 for now. - order on decrement - acq_rel is not always
>         needed, I believe it is. - corner cases where the decrement is
>         known to not reach zero, Then don't use the helper in that
>         case? - corner cases where the decrement cannot be permitted
>         to reach zero(e.g. vlc_object_t, IIRC). Ditto. -- Envoyé de
>         mon appareil Android avec Courriel K-9 Mail. Veuillez excuser
>         ma brièveté.
>         ------------------------------------------------------------------------
>         vlc-devel mailing list To unsubscribe or modify your
>         subscription options:
>         https://mailman.videolan.org/listinfo/vlc-devel
>         ------------------------------------------------------------------------
>         vlc-devel mailing list To unsubscribe or modify your
>         subscription options:
>         https://mailman.videolan.org/listinfo/vlc-devel -- Envoyé de
>         mon appareil Android avec Courriel K-9 Mail. Veuillez excuser
>         ma brièveté.
>         ------------------------------------------------------------------------
>         vlc-devel mailing list To unsubscribe or modify your
>         subscription options:
>         https://mailman.videolan.org/listinfo/vlc-devel 
>
>
>     ------------------------------------------------------------------------
>
>     vlc-devel mailing list
>     To unsubscribe or modify your subscription options:
>     https://mailman.videolan.org/listinfo/vlc-devel
>
>
> -- 
> Envoyé de mon appareil Android avec Courriel K-9 Mail. Veuillez 
> excuser ma brièveté.
>
>
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel