Closed mundya closed 8 years ago
@mossblaser - the coverage seems to indicate a problem with stuff touched by #217 rather than by this PR.
Superficial criticisms aside this looks good to me.
The coverage issue (as mentioned offline) appears to be a result of coveralls doing something a bit strange and I'll investigate at some other time. As such I'm happy to ignore it for this PR since we've both verified the coverage locally is actually still 100%.
I've retained the brackets in
not (key & bit)
because I feel it's clearer thannot key & bit
.
I would agree :).
LGTM! Merge at will once travis is happy.
... I must figure out what's going on with coveralls at some point :/
@rowleya / @alan-stokes - this should improve the compression of the ucolumn tables.
Insert entries above blocks of entries of equivalent generality. This makes that the Python implementation of OC matches (a) the C version for SpiNNaker (b) that described in the HPSR submission.
For table ordering, see https://github.com/project-rig/rig_routing_tables/blob/master/include/ordered_covering.h#L16-L57
For downcheck preferring to modify more significant bits in merges see https://github.com/project-rig/rig_routing_tables/blob/master/include/ordered_covering.h#L285-L292
For merge application, see https://github.com/project-rig/rig_routing_tables/blob/master/include/ordered_covering.h#L409-L485