anyNA requires R >= 3.1.0

[mjkallen]

@jeroenooms you added the anyNA() function in commit ff76a61 but you don't require R >= 3.1.0 in the DESCRIPTION file. I get an error that anyNA() is not found when using R < 3.1.0. Either you do the sensible thing and just use any(is.na(keys)) or you bump the minimum required version of R to 3.1.0.

[jeroen]

Fixed @ 5e38037. Why are you using a deprecated version of R?

[jeroen]

If you find any other compatibility problems with pre r-oldrel versions just send a PR :)

[mjkallen]

Thanks for the fix and, for good order,

1. I wouldn't consider any 3.x version of R as deprecated, even if it's pre r-oldrel and if you do want to consider these deprecated then you should just depend on R (>=3.1.3) in the DESCRIPTION file,
2. using anyNA() is unnecessary as keys is not a huge vector on which my computer is going to choke,
3. when people started using that other useless function paste0 they correctly bumped their dependency to R (>=2.15.0) so their is a precedent for these cases.

[jeroen]

I don't mind at all, but it is difficult to detect these kind of problems because the CRAN testing suite only tests on r-oldrel, r-release and r-devel. I'm sure there are other things in the package that require at least 3.0 or so, but it would be silly to check when each function was introduced. Afaik the convention is to only declare a minimum version if it requires something more recent than r-oldrel at the time of the release of the package.

You didn't answer the question by the way :) Does renjin target the 3.0 reference implementation?

[mjkallen]

Renjin is currently a fork of R 2.14.2, but I was using R 3.0.2 in a completely different context when I hit this error.

Speaking of paste0: you use it in this package as well so R (>=2.10.0) is not correct, even with your latest fix.