Re: windowscodecs: Disable ICNS support when building for Mac OS X 10.4 Tiger.
Should we reconsider libicns? Apple's APIs obviously can't be relied upon. On Fri, Jan 21, 2011 at 2:58 AM, Ken Thomases <ken(a)codeweavers.com> wrote:
For http://bugs.winehq.org/show_bug.cgi?id=25778 --- dlls/windowscodecs/icnsformat.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-)
On 21 January 2011 08:36, Damjan Jovanovic <damjan.jov(a)gmail.com> wrote:
Should we reconsider libicns? Apple's APIs obviously can't be relied upon.
You could always: 1. check for native support (including presence of kIconServices16PixelDataARGB) and use if present; 2. check for libicns and use it if present; 3. disable ICNS support. using autoconf checks. That way, support is enabled where available (instead of relying on OS version detection logic). NOTE: the libicns check is optional as it would require maintaining both native OSX and libicns versions. - Reece
On Fri, Jan 21, 2011 at 10:52 AM, Reece Dunn <msclrhd(a)googlemail.com> wrote:
On 21 January 2011 08:36, Damjan Jovanovic <damjan.jov(a)gmail.com> wrote:
Should we reconsider libicns? Apple's APIs obviously can't be relied upon.
You could always:
1. check for native support (including presence of kIconServices16PixelDataARGB) and use if present; 2. check for libicns and use it if present; 3. disable ICNS support.
using autoconf checks. That way, support is enabled where available (instead of relying on OS version detection logic).
NOTE: the libicns check is optional as it would require maintaining both native OSX and libicns versions.
- Reece
About option 3: December 2010 desktop OS market share, Wikimedia stats: Intel MacOS X 10.4 0.49% Ubuntu 10.10 0.25% Ubuntu 10.04 0.34%. If distribution of Wine users among Linux and MacOS is roughly equal, then even if we disable icons on only Intel MacOS X 10.4, it will break for almost as many users as are using Ubuntu 10.04 and 10.10 put together. Damjan
Damjan Jovanovic <damjan.jov(a)gmail.com> writes:
If distribution of Wine users among Linux and MacOS is roughly equal, then even if we disable icons on only Intel MacOS X 10.4, it will break for almost as many users as are using Ubuntu 10.04 and 10.10 put together.
Several other things are already broken on Mac OS X 10.4. Basic functionality is there, but it's not worth investing a lot of efforts into supporting every last bit of functionality on 10.4. If you want the latest and greatest Wine I don't think it's unreasonable to ask that you upgrade the rest of the OS too. -- Alexandre Julliard julliard(a)winehq.org
On Jan 21, 2011, at 2:36 AM, Damjan Jovanovic wrote:
Should we reconsider libicns? Apple's APIs obviously can't be relied upon.
On Jan 21, 2011, at 5:04 AM, Alexandre Julliard wrote:
Several other things are already broken on Mac OS X 10.4. Basic functionality is there, but it's not worth investing a lot of efforts into supporting every last bit of functionality on 10.4.
In addition to what Alexandre said, if we did want to put in the effort to maintain Tiger support, we'd still use the native API. Tiger does have support for ICNS and the IconFamily API; that API goes back to classic Mac OS. It's just that Leopard introduced some conveniences to that API -- in particular, treating sizes smaller than 256 as ARGB instead of separate image and mask -- that I took advantage of in my code. So, supporting Tiger would entail more cumbersome code, but not a reversion to libicns. Regards, Ken
participants (4)
-
Alexandre Julliard -
Damjan Jovanovic -
Ken Thomases -
Reece Dunn