[Adium-devl] 10.5

[Ryan Govostes]

Though I'm completely for dropping 10.4 soon, I wouldn't want to spend  
a lot of time updating code that will need to be changed yet again for  
10.6 -- assuming it's a free or near-free update. But I don't want  
this to end up like it did last year [1].


On Aug 9, 2008, at 6:27 PM, David Smith wrote:

> 	Use fast enumeration (slight speed and safety improvements, more
> concise and readable code)

Some stateless enumerations would probably be updated again for 10.6  
to make use of Grand Central.

> 	Use xcconfig files (Makes managing project configuration
> significantly simpler)

Quicksilver is currently using these and it makes the project much  
nicer to work with. Considering my .pbproj file gets messed up every  
once in a while, it'd be a welcome change.

> 	GCC 4.2/GCC-LLVM (not sure if there's any benefit for us here. 4.2
> might have some better warnings and such)

Looking over the changes to gcc-4.1 [3], we'd gain a handful of  
optimizations. Hopefully gcc-4.3 will reach Apple developers soon so  
we can take advantage of those changes [4] too. Using gcc-llvm gains  
us nothing in my unscientific tests.


> Options:
>
> 1) Keep 1.3 as a semi-active branch while 1.4 goes Leopard-only.
> MSNP15, Sparkle 1.5, etc... would bake in trunk for a while, then
> merge to 1.3.x. I think I'm currently in favor of this plan.

Dimmuxx and forum users have been testing MSNP15 builds for a little  
while, so we can hit the ground running here.

Sparkle 1.5 doesn't seem like it would need to "bake," but I'll defer  
to someone who knows more about it than I do.

> 3) As 2, except with a 1.5 branch active in parallel.

> 	Upsides: Avoids the downsides of 2, and the stability concerns of 1
> 	Downsides: Have we ever been able to handle 3 active branches? Not
> that I'm aware of.

Since 1.4 will not (I assume) require too much activity, I favor this.  
We can shift the tickets between the two releases so that 1.4 gets  
released as soon as we feel it's stable.

Plus we'll become experts in svn merge :-)

> 4) Just drop Tiger in 1.4 and proceed as normal.
> 	Upsides: Easy
> 	Downsides: Leaves a good chunk (maybe 20% at release?) of our
> userbase unable to upgrade to the latest version

The users who don't upgrade their OS have no leg to stand on if they  
complain about being unable to upgrade their IM client. They've  
already established they're willing to use outdated software.

Regards,
Ryan Govostes

[1] http://rgov.org/catch-22-nda-vs-gpl
[2] http://trac.adiumx.com/ticket/10308
[3] http://gcc.gnu.org/gcc-4.1/changes.html
[4] http://gcc.gnu.org/gcc-4.3/changes.html