aserio changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/
parsa has joined #ste||ar
<heller>
hkaiser: I guess we are totally not on the same page ;)
<heller>
hkaiser: all I am trying to say is that that !localities_ case was encoded in the condition localities_.empty() (prior to your PR). Now localities_ should never be empty, and the empty case is coerced into the case where the size of the vector is 1
<heller>
and now, in the "empty" case, we create a shared pointer of a vector holding one id_type just to destroy right away after we checked that it's "empty"
<heller>
I might miss something ...
<hkaiser>
heller: I think th eold behavior was not correct
<heller>
ok
<heller>
the tests we have seem to pass
<github>
[hpx] hkaiser force-pushed fixing_3054 from 627b901 to ecf6b6f: https://git.io/vb5lI
<github>
hpx/fixing_3054 ecf6b6f Hartmut Kaiser: Unify handling of list of localities...
<hkaiser>
shrug, we don't have any thorough tests for this policy
<heller>
poilcies in general are not heavily used, i guess
<hkaiser>
well, num_partitions_ could be zero, I guess
<heller>
hmm
<heller>
ok
<heller>
doesn't hurt
<heller>
here is the thing that I don't like about the PR: In the case of a default constructed policy, ensure_localities is being called over and over again
<hkaiser>
no
<hkaiser>
only for the container_policy
<hkaiser>
I removed it completly from the default_policy
<heller>
ahh, you changed it back, great
<heller>
now we are on the same page ;)
<hkaiser>
I can remove it from one more spot
<github>
[hpx] hkaiser force-pushed fixing_3054 from ecf6b6f to 0995a14: https://git.io/vb5lI
<github>
hpx/fixing_3054 0995a14 Hartmut Kaiser: Unify handling of list of localities...