[petsc-dev] clang is having some problems with elemental

[Barry Smith]

  I agree that sometimes Apple makes bad decisions on their development systems. But I do find them usable, unlike Microsofts which are completely unusable if you want to maintain portability with anything else.


On Dec 4, 2013, at 11:45 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

> Barry Smith <bsmith at mcs.anl.gov> writes:
>>   And the reason is always in some way related to $
> 
> As with all decisions (time is money).  Jack chose to require the latest
> compilers because it saves him development time.  PETSc supports ancient
> compilers, MPI implementations, and Python because we think that the
> time our users save by not having to upgrade their systems (even if we
> think their reasons for not upgrading are bogus) is more important than
> our time required to maintain that compatibility.
> 
> Apple continues to not assign a developer to make gdb and valgrind work
> as well on Apple systems as on Linux/BSD systems because (evidently)
> they think that the pain they inflict on the developers will not cause
> those people to stop buying Apple products.  (I blame Apple for this
> state of affairs because they take a working product and change
> non-broken things in half-assed ways, like -framework and dsymutil, in
> their highly-profitable proprietary system, then expect the developers
> of those open source tools to volunteer their time and purchase Apple
> hardware to maintain compatibility.)