JSS prefers to publish papers about packages where the package is on CRAN and has been there long enough to have matured (i.e., obvious bugs ironed out and a few active users). This is partly because we have so many submissions that it helps to filter some out and this approach provides some basic quality checks. So I suggest you begin by developing the package for CRAN. This is a preference rather than a requirement, and it is not stated anywhere in the JSS rules. A paper describing a package that has only recently been put on CRAN will still be considered, but the probability of it getting through the reviewing process is smaller.
(http://robjhyndman.com/hyndsight/jss-rpackages/)
JSS prefers to publish papers about packages where the package is on CRAN and has been there long enough to have matured (i.e., obvious bugs ironed out and a few active users). This is partly because we have so many submissions that it helps to filter some out and this approach provides some basic quality checks. So I suggest you begin by developing the package for CRAN. This is a preference rather than a requirement, and it is not stated anywhere in the JSS rules. A paper describing a package that has only recently been put on CRAN will still be considered, but the probability of it getting through the reviewing process is smaller. (http://robjhyndman.com/hyndsight/jss-rpackages/)