Please make Pre-3.1 read.csv (type.convert) behavior available
Hi Dirk, Thanks for taking the time to respond (both here and in other forums). Most of what I wanted to share I put in a followup response to Duncan (please read that thread if you?re interested). I would like to comment on the last point you brought up, though, in case anyone else finds it beneficial. For data which is exchanged programmatically machine-to-machine, I was able to use Java?s Double.toHexString() as a direct replacement for toString(). R is able to read this lossless (but still text) format. So this addresses some of the challenges we have with this change. Thanks, Tom
On Apr 26, 2014, at 5:26 AM, Dirk Eddelbuettel <edd at debian.org> wrote:
On 26 April 2014 at 07:28, Duncan Murdoch wrote: | On 26/04/2014, 12:23 AM, Tom Kraljevic wrote: | > | > Hi, | > | > We at 0xdata use Java and R together, and the new behavior for read.csv has | > made R unable to read the output of Java?s Double.toString(). | | It may be less convenient, but it's certainly not "unable". Use colClasses. | | | > | > This, needless to say, is disruptive for us. (Actually, it was downright shocking.) | | It wouldn't have been a shock if you had tested pre-release versions. | Commercial users of R should be contributing to its development, and | that's a really easy way to do so. Seconded. For what it is worth, I made five pre-release available within Debian. Testing thses each was just an apt-get away. In any event, you can also farm out the old behaviour to a (local or even CRAN) package that provides the old behaviour if your life depends upon it. Or you could real serialization rather than relying on the crutch that is csv. Dirk -- Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com