Ok, make sense.

The alternative syntax would here be to use arange from numpy.

b2 = b[arange(0, b.local_size(), 2)]

Johan


On Fri, Dec 12, 2014 at 4:20 PM, Martin Sandve Alnæs <[email protected]>
wrote:
>
> The policy is to avoid it if possible. However I think this is a good case.
> It's a while since 1.0 now, some things will have to change over time.
>
> Just keep the line commented out, and add a comment like
> # This feature has been removed because it is unsafe in parallel:
> or something like that. Even better would be to add an alternative
> formulation, but I don't think that line makes much sense anyway?
>
> Martin
>
>
> On 12 December 2014 at 16:13, Johan Hake <[email protected]> wrote:
>
>> Hello!
>>
>> I am about to push the numpy index access for GenericVectors. As
>> previously discussed I will remove support for slicing access. This breaks
>> one book example. What is the policy for that? That was the whole point of
>> having these tests there.
>>
>> def test_p51_box_2():
>>     b = Vector(mpi_comm_world(), 10)
>>     c = Vector(mpi_comm_world(), 10)
>>     b_copy = b[:]
>>     b[:] = c
>>     b[b < 0] = 0
>>     b2 = b[::2]
>>
>> The last line wont work.
>>
>> Johan
>>
>> _______________________________________________
>> fenics mailing list
>> [email protected]
>> http://fenicsproject.org/mailman/listinfo/fenics
>>
>>
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics

Reply via email to