Sunday, April 4, 2010
I see there have been some problems in getting files down and running, so here are some suggestions that may help.
Text version of download
To download direct from the repository page, the best way may be cut and paste - other ways may lead to HTML issues. But best may be to use the document repository. I'd suggest just downloading the All-R.zip - it's very small. The zipfile has a directory structure that collects files related to blog posts. If you download All-Files.zip you'll get the same directory structure, but with results and anything else that seemed relevant.
Input files for R
Here I haven't been systematic enough. Most R files call input files with a scan() or a read.fwf(). I've sometimes referenced the files with aspects of my directory structure, whereas of course they should reflect your structure.
I'll try to make sure to reference the input files as if in the current directory. But I'll also put at the start of the program a set of commands saying
These will list the input files. You should insert your own paths and filenames (if different)
Update I'll extend this idea to putting a section near the top where all the variables (including input files) )that you might want to change are initialised. Instead of loc=, for input files I'll write "input1 = " etc
Text conventions for repository
I have tried to separate R code from my text with font changes (italic). But I'll try to improve by marking the beginning and end with comment lines
######## Start of prog.R ###########
######## End of prog.R ###########
The name is just a suggestion 0 if you down load from there you'll have to supply a name, which can be anything you like. But it probably corresponds to the name in the doc drop.
I'll try to put readme files in the doc repository and the zip files. These should describe where the files belong.
The blog post should tell you where to find the input files needed. GHCN files for example come from http://www1.ncdc.noaa.gov/pub/data/ghcn/v2/
Rememner to uncompress files before passing to R.
I've heard of trouble in lodging comments. The settings, as I understand them, should provide no restrictions. Let me know if you can't get through.
Blogspot makes pesky issues for commenting without identity. You won't be able to paste, for example. But if you use any identity (Google, Typekey, Wordpress etc) it's much better. This may also help if comments aren't being accepted.
If all else fails
My email name is anstokes, and you can find me at westnet + com + au
Posted by Nick Stokes at 8:51 PM