-
The randomisations should have a rename option.
-
profiling shows some bottlenecks in Biodiverse::Randomise::rand_structured. These need to be opened up.
-
```
When a user runs a randomisation the system currently desensitises most of
the setup widgets (those that should not be changed again for this
randomisation).
It should also do any of the method s…
-
```
The randomisations currently do nothing to the group properties, and simply
re-assign them to the randomised basedata.
An option needs to be added to control this behaviour.
The same goes f…
-
Possible solution to the bottleneck flagged in #685 is at https://stackoverflow.com/a/5684892
This would avoid the need for binary searches across sorted lists, providing the insertion and deletion…
-
Benchmarking shows that the List::MoreUtils binary search functions are faster than the current implementation using List::BinarySearch, probably due to subroutine overheads. The LMU code is entirely…
-
The "checkpoint save iterations" option in the randomisations is from back when the memory leaks in issue #5 were causing grief.
This should be off by default, for which the simplest approach is to…
-
It would be useful to add a randomisation where the labels are randomly assigned within non-overlapping subsets rather than across the entire basedata.
That way one could, for example, randomise th…
-
```
Existing randomisations override any user specified analysis arguments. This
is currently done at the level of the randomisation method, rather than at the
run_analysis level.
This has worke…
-
There needs to be the capacity to reintegrate cloned basedatas after randomisations have been run in parallel.
Key things to ensure are that the basedatas
1. Have the same outputs. Can check nam…