[Libreoffice] Mac OS 10.4 Support

[Jonathan Aquilina]

On 25/07/2011 03:24, Christian Lohmaier wrote:
> Hi Norbert, *,
>
> On Mon, Jul 25, 2011 at 2:52 AM, Norbert Thiebaud<nthiebaud at gmail.com>  wrote:
>> On Sun, Jul 24, 2011 at 7:10 PM, Christian Lohmaier
>> <lohmaier+libreoffice at googlemail.com>  wrote:
>>>> because you can build with a sdk 10.6 and a deployment-target of 10.4
>>>> (at least that is how I understood the<AvailabilityMacros.h>  )
>>> Yes, in theory, but see my other post, or look at the apple developer
>>> lists. That it only works in theory is the reason why people also
>>> prefer to keep using the 3.2.6 SDK even on system where you could be
>>> using the newer 4.x ones.
>> I'm getting a bit confuse. --with-macosx-target-deployment is aiming
> --with-macosx-deployment-target
>
>> at MACOSX_DEPLOYMENT_TARGET right ?
> Yes, but..
>
>> That is not solid enough for the reason you stated above. hence the
>> --with-mac-sdk= which force the use of a particular version of the SDK
> it should also set the matching SDK.
>
>> I mean, we could use --with-macos-target-deployment to do the task I
>> assigned to --with-mac-sdk.. but that would certainly not  be more
>> 'consistent'
> It is. As you do expect a version string as argument passed to that
> option, not a path to the actual SDK.
>
> --with-macosx-deployment-target=10.4
>
> it's clear that you want the build to run on 10.4 and later.
>
> --with-macosx-sdk=10.6 (or /Developer/SDKs...10.6, doesn't matter)
>
> it could mean: "I only have XCode 4.1 installed, so I don't have the
> 10.4 SDK available and cannot use it, but I still would like to have
> my build run on my laptop that runs 10.4"
>
> Just specifying the sdk is ambiguous. As you yourself noted that it is
> not strictly necessary (in theory) to use the target's SDK to produce
> a build that runs on the target.
>
> If people consider it a problem to modify the XCode 3.2.6 dmg to be
> able to install it on Lion, then using the "flaky" way to rely on the
> availability macros is the only way to keep 10.4 and 10.5 compatible.
>
> The other way doesn't make any sense, you cannot compile against 10.4
> SDK but make use of 10.5 features (and it doesn't make sense to use
> the 10.7SDK to compile for 10.4 or 10.5, this is just asking for
> deprecation warnings and stuff) So IMHO one switch that specifies the
> target is enough to derive the rest.
>
> If matching SDK is available, use that, otherwise use whatever
> "oldest" SDK is available and hope for the availability-macro stuff
> (and more importantly the linking) doesn't fail.
> This would of course require some more modification to the compiler /
> linker arguments to support that method.

 From what I have seen with mac package, how come we dont package a 
universal binary that will support older hardware as well as the new 
intel based mac's

[SNIP]