Add non-member Kokkos::resize/realloc for DualView

[AndrewGaspar]

It would be nice if there were non-member Kokkos::resize and realloc functions to match Kokkos::View.

Slack discussion:

Andrew Gaspar [10:50 AM]
Could we get an overload of Kokkos::resize for DualView?

crtrott [10:54 AM]
It doesnt exist?

Andrew Gaspar [10:54 AM]
There’s member resize

crtrott [10:54 AM]
Oh maybe it got a member functiin
Yeah

Andrew Gaspar [10:55 AM]
but it would be nice for a couple reasons if it was non-member resize like View:

1. If you change a View to a DualView, you don’t have to change any resizes
2. Visually consistent in the code

crtrott [10:56 AM]
Agreed

Andrew Gaspar [10:56 AM]
I can file an issue

crtrott [10:56 AM]
Jup do that

Andrew Gaspar [10:57 AM]
I can see about providing the change myself

crtrott [10:57 AM]
Also realloc

[dsunder]

There are kokkos_malloc, kokkos_free, and kokkos_realloc available as free functions in the Kokkos namespace, though they work on raw allocations and not Kokkos data structures. Kokkos::resize is possible for View, DualView, and UnorderedMap, not sure how realloc would be different from resize (unless realloc does not copy the data). Cheers,

…

--Dan S From: Andrew Gaspar <notifications@github.com> Reply-To: kokkos/kokkos <reply@reply.github.com> Date: Wednesday, June 5, 2019 at 10:58 AM To: kokkos/kokkos <kokkos@noreply.github.com> Cc: Subscribed <subscribed@noreply.github.com> Subject: [EXTERNAL] [kokkos/kokkos] Add non-member Kokkos::resize/realloc for DualView (#2170) It would be nice if there were non-member Kokkos::resize and realloc functions to match Kokkos::View. Slack discussion: Andrew Gaspar [10:50 AM] Could we get an overload of Kokkos::resize for DualView? crtrott [10:54 AM] It doesnt exist? Andrew Gaspar [10:54 AM] There’s member resize crtrott [10:54 AM] Oh maybe it got a member functiin Yeah Andrew Gaspar [10:55 AM] but it would be nice for a couple reasons if it was non-member resize like View: 1. If you change a View to a DualView, you don’t have to change any resizes 2. Visually consistent in the code crtrott [10:56 AM] Agreed Andrew Gaspar [10:56 AM] I can file an issue crtrott [10:56 AM] Jup do that Andrew Gaspar [10:57 AM] I can see about providing the change myself crtrott [10:57 AM] Also realloc — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#2170?email_source=notifications&email_token=ACBQOEVJSWAMBNQXVCTHLNDPY7WEPA5CNFSM4HT7P7SKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GX2KLUQ>, or mute the thread<https://github.com/notifications/unsubscribe-auth/ACBQOEQDIIGR243GSBPF4L3PY7WEPANCNFSM4HT7P7SA>.

[stanmoore1]

Yes resize preserves the existing data of the view while realloc does not.

[dhollman]

@AndrewGaspar does PR #2408 address this?

[AndrewGaspar]

Looks right.