On Mon, Mar 30, 2015 at 4:55 AM, RenĂ© J.V. <[email protected]> wrote:
> On Monday March 30 2015 01:40:39 Jeremy Huddleston Sequoia wrote:
>>Actually, I think I'll just do similarly and revbump all dependents.  
>>Unfortunately, that means anyone using the library for their own projects 
>>will need to recompile, but that's probably better than the maintenance 
>>headache of being different than upstream going forward.  Hopefully they can 
>>manage to version themselves correctly now that they're on cmake.
>
> I'm guessing this is about the  compatibility and min/max versioning stored 
> in the dylib, so providing a transition period would mean copying dylib(s) 
> rather than symlinking to them? Still, if you copy only .N.dylib variants 
> that are stored in dependent binaries (and change their internal versioning 
> info) that should satisfy existing binaries until they're recompiled, no?
>
> @MLD: we could probably pull the same trick with the port where we bump down 
> the compatibility version, no? (qca? Can't remember which one it is ATM..)
>
> R.

The 'problem' is that cmake automates the setting of the current and
compatibility versions from the VERSION set in CMakeLists.txt. There
is no fine control in cmake for setting a compatibility version that
is different from VERSION (which potentially might be worth requesting
as an enhancement to cmake upstream).

> _______________________________________________
> macports-dev mailing list
> [email protected]
> https://lists.macosforge.org/mailman/listinfo/macports-dev
_______________________________________________
macports-dev mailing list
[email protected]
https://lists.macosforge.org/mailman/listinfo/macports-dev

Reply via email to