Added function to report Annoy library version.

[LTLA]

As briefly mentioned in jlmelville/uwot#69. We can do:

library(RcppAnnoy)
version()
## major minor patch devel 
##     0     0    17     2 

The motivation: BiocNeighbors will vendor RcppAnnoy"s headers, but the former's unit tests still compare its search results to those from RcppAnnoy's R interface (just to check that I wrote the encapsulating C++ code correctly). These search results are only expected to be equal when the Annoy libraries are of the same version, otherwise the tests should be skipped. We can check this easily with these version() functions, both in RcppAnnoy and plus a counterpart in BiocNeighbors.

P.S. Not sure why line 4 appeared in RcppExports.cpp, that was just automatically generated.

[eddelbuettel]

Version accessor is good, will merge. Line 4 is autogenerated by Rcpp Attributes. Not sure why it wasn't there before.

The one thing I hate hate HATE hate with a vengeance is your indentation. That code was meant to be written as

Rcpp::IntegerVector version () {
    return Rcpp::IntegerVector::create(Rcpp::Named("major")=RCPPANNOY_VERSION_MAJOR,
                                       Rcpp::Named("minor")=RCPPANNOY_VERSION_MINOR,
                                       Rcpp::Named("patch")=RCPPANNOY_VERSION_PATCH,
                                       Rcpp::Named("devel")=RCPPANNOY_VERSION_MICRO);
}

But I pretend I never saw your version and move on. Life is too short to argue over whitespace. If we cared we'd be using Python already...

So are you set then for your testing?

[eddelbuettel]

Shall we call it annoy_version() though? I do the same here

> tiledb::tiledb_version(TRUE)
[1] ‘2.2.0’
> tiledb::tiledb_version()
major minor patch 
    2     2     0 
> 

[LTLA]

Ha! Well, when in Rome...

Do you want camelCase or snake_case? I notice that the architecture getter is camelCase, so maybe getAnnoyVersion()?

[eddelbuettel]

(Nerdyness aside I am mostly curious what people use for editors. What is your default? I think I have seen that awful scheme too via clang-format but I figured you wouldn't run that locally).

I used to used camelCase quite religiously and mostly exclusively and seem to slowly be giving in to snake_case. But as you say: "when in Rome..." Everything that is exported in that package is camelCase though so getAnnoyVersion() sounds good in this context.

[eddelbuettel]

Not sure why it wasn't there before.

Probably because we added a header conforming to the pattern it looks for, and I simply had not run compileAttributes().

[LTLA]

Done, mimicking tiledb_version's functionality.

(Just vim with auto/smart-indenting turned on. I think the exact behavior of the indenter changes with whatever version of vim I happen to get on my computer; so you could classify my lines of code based on the computer I wrote it on!)

[eddelbuettel]

You even changed the indentation! Next beverage on me if we ever get to meet post pandemic...