reorder passed info when setting new pop and strat

[thibautjombart]

If foo is named in these operations:

pop(x) <- foo
strata(x) <- foo

then it would make sense to check the names against indNames(x) and:

1. issue error/warning if mismatches
2. reorder otherwise

@zkamvar what do you think?

[romunov]

How about preserving the pop(x) order of names and match names and assign values? Would this warrant a warning or a note?

[zkamvar]

This could definitely be possible. @romunov, I don't think it would be wise to change the order of the data to mach the newly-assigned pop.

I DO think it would be a good idea to add the indNames to the unnamed vectors.

I also think think it would be a good idea to issue a warning if the population or the strata were in the wrong order.

• check for named vector/data.frames
• re-order and warn if necessary
• name vectors that are unnamed
• add documentation in the tutorials/documentation.

[romunov]

I would also add a unit test for this.

[zkamvar]

good point! - [ ] unit tests

[zkamvar]

One more thing that should be considered with this:

Some monte carlo tests involve shuffling the data by using:

pop(x) <- sample(pop(x))

Under the proposed changes, this would do absolutely nothing, which would effectively break the above procedure. The remedy for this is to recommend users to do the following with monte carlo tests:

pop(x) <- unname(sample(pop(x)))

Perhaps one way to get around this would be to add a check.names = FALSE argument to the pop() strata() and other() constructors, recommending that people use:

pop(x, check.names = TRUE) <- myPop

when setting a new population? I'm still a bit on the fence about it because, on the one hand, you don't break things, but on the other hand, it's harder to enforce good practices. Additionally, there is something slightly unsettling about having an unordered population factor floating around.... but then again, there could be a method called check() for the genind which will reset the population factor/strata/other if they are named....

[thibautjombart]

Yes I think we want to keep the default behaviour of pop <- so that pop(x) <- sample(pop(x)) still does the same; adding a check.name argument would make sense. How about a second argument reorder which would trigger the automated reordering, defaulting to check.names? The use case would look like:

pop(x) <- sample(pop(x)) # re-shuffle pop
pop(x, check.names = TRUE) <- myPop # assign new pop with automated matching
pop(x, check.names = TRUE, reorder = FALSE) <- myPop # merely a sanity check

I am not sure if we want the last use case. If not, I would rename the check.name into something more explicit, e.g. match.labels or match.names.