Open barracuda156 opened 1 year ago
thanks for the report - hopefully I'll get some time to work on xtools later in the year, I guess this should be fixable.
P. S. By the way, libkern/version.h
inside the framework (unfortunately, lacking in /usr/include
) has all needed defines to differentiate between released and various stages of developer builds; it should be rather easy to fix 10A190 in way that will not affect 10.6.x Rosetta and therefore can be mergeable to upstream. I am just not sure how to include it in a neat way without specifying complete path /System/Library/Frameworks/Kernel.framework/Versions/A/Headers/libkern/version.h
inside a define of OS version.
If included into gcc/config/rs6000/darwin.h
, a fallback for 10A190 can be included inside #if version_stage == 0x20
condition, like it has now fallbacks for Rosetta.
This is what 10.6.8 has:
/* VERSION_STAGE, version_stage, is an integer set to one of the following: */
#define VERSION_STAGE_DEV 0x20
#define VERSION_STAGE_ALPHA 0x40
#define VERSION_STAGE_BETA 0x60
#define VERSION_STAGE_RELEASE 0x80
#define VERSION_STAGE VERSION_STAGE_RELEASE
extern const int version_stage;
10A190 has VERSION_STAGE defined to VERSION_STAGE_DEV.
@iains At the same time Apple ld64-97 works fine. Maybe it is possible to unbreak this? I am interested to test xtools, and perhaps we gain nothing by not recognizing developer versions. (This is not even a PPC issue, there are far more x86 developer builds around.)