Rcpp exports duplicated

[johnlaing]

To resolve #405 we now define every exported C++ function twice. This is fine because preprocessing, but we should only Rcpp::export each function once - else a call to compileAttributes() results in duplicate export entries.

[eddelbuettel]

Nice catch. It so happens that I have a fix (I think) as part of another cleanup / rearrangement of the conditional compilation at also makes it easier to follow it. I can commit, but I was waiting for you to (if at all possible) get to PR #404 first?

[johnlaing]

I was looking at #404 when I stumbled across this. The fix is trivial (drop the extra exports tags!) and I will submit PR shortly.

[eddelbuettel]

We can do it the other way around too: I could clean this up, and we rebase #404 afterwards.

The basic idea of what I have tested is to

• pass one overall 'yes/no' from configure to the Makefile and have the Makefile simplified as it can then have if/else
• with that if becomes easy to define this in terms of a 'yes/no' variable that is flipped to 'yes'
• and have compilation of parts interfacing blp on for the yes

But if you want you could write a PR now too I suppose under the existing structure and have the #if !defined(...) inside the function. Thinko of mine to not have done so when the stanza where I used this before does it too -- and likely for the same reason!

[eddelbuettel]

This has been taken care of in #409