Re: [Qemu-devel] RFC: delete darwin-user ?

[Alexander Graf]

On 13.04.2012, at 02:01, Andreas Färber wrote:

> Am 13.04.2012 01:57, schrieb Peter Maydell:
>> So, the darwin-user/ subdirectory is officially Orphan, and looking
>> at the git history the last commit to it which wasn't either "apply
>> a global change", "fix cppcheck/spelling error/similar autodetected
>> nit" or "duplicate a linux-user fix into darwin-user and bsd-user"
>> was way back in 2009, perhaps even earlier.
>> 
>> If in three years nobody's cared enough about it to manage to get
>> a single darwin-user specific change into the tree, my suggestion
>> would be that it is in practice dead and we should make that
>> official by deleting it. Then we can stop wasting effort on applying
>> (untested!) global changes to these files.
>> 
>> Does anybody want to argue against that idea?
>> 
>> (This came up on IRC a month or so ago, and there seemed to be
>> a consensus that the use case it was originally intended for
>> (x86 darwin binaries on ppc or vice versa? I forget) was now
>> no longer interesting to anybody. I don't have a strong opinion
>> personally but I thought I'd bring the idea up formally here.)
> 
> I have an RFC series coming up that cleans up MAINTAINERS and does this
> along the way.
> 
> I was the one to bring this up on IRC and I had been previously in favor
> of keeping darwin-user with Apple's drop of Rosetta (ppc-on-x86) in
> mind. Mac OS X Lion came, dropped Rosetta and no one sent any patch, so
> I am now in favor of dropping it because I can't test the CPU changes
> I'm doing.

I'm also in full favor of dropping it. The syscall interface on Darwin is 
stable on the libSystem layer, not the syscall layer, so whatever is there 
won't work with anything recent anyways :).


Alex