| Re: [AD] set_config_file jumping around on OS X bundles |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
On Monday 22 August 2005 22:28, Matthew Leverton wrote:
> "Once Allegro 4.2 is released, we plan to maintain *backward*
> compatibility at the Application Binary Interface level for the
> subsequent releases of the 4.2.x series. For example, that means you
> will be able to use an executable compiled using version 4.2.0 with
> version 4.2.5 or 4.2.41 of the dynamically linked library"
>
> It doesn't say that a 4.2.5 games works with 4.2.0 DLL. Nor should it.
> The 4.0 terminology specifically mentioned backward and forward
> compatibility via examples. The word "backward" was added to the 4.2
> docs, and the example of forward compatibility was removed.
>
> Forward compatibility: 4.2.5 games will work on the 4.2.0 DLL.
> Backward compatibility: 4.2.0 games will work on the 4.2.5 DLL.
>
> I'm confused on whether or not any change of policy has taken place
> for the 4.2.0 series. I know at one point, only backward compatibility
> was being considered, but now is forward as well?
I think we agreed to stick to backward compatibility.
That still means a game compiled against 4.2.0 must work with 4.2.5 though,
which still means we can't add things to vtables. This changes the size of
the structs and may change function pointers as well.
As I said, we can work around this problem for individual cases by using
#ifdef PLATFORM / #endif code. Not pretty, but it does work well enough.
> >Considering that 4.3 will be a major update, probably one the biggest
> in Allegro
> >history I don't think 4.4 can be released so soon.
> >
> Yes, that is what concerns me as well.
Yes, I don't think it's a good idea to just slap a `stable' tag on a WIP
release in 4.3 and continue with 4.5... what do we mean then by `stable'?
I see 4.4 coming out when at least a part of the API has changed or when
the entire library is prefixed. Either way, I think a year is optimistic.
Evert