[R] Reading sas7bdat files directly

[Frank E Harrell Jr]

Chris Long wrote:
> No problem, Frank, I'm glad that you think it will be useful.
> 
> I will be making changes to dsread in the coming weeks so you may want to
> hold off with your helper function in case my changes break it (the
> formatting of the variable metadata listing may well change).  Re: the
> metadata, feel free to suggest an alternate output format that would make
> this a cleaner process.  For example, I have in mind to allow the -c
> (contents) and -v (CSV) flags together, in which case the dataset metadata
> will be output in CSV format.
> 
> Any other suggestions or comments are welcomed.
> 
> Chris.

Thanks very much for your note Chris.  Having an option to output the 
metadata in a standard csv format as well as the nice table format you 
provide already (either one or the other) will be good.  If you allow 
flags (one could be -o) to specify the names of 2 different output files 
when STDOUT is not being used, that will expedite things.

As sas7bdat files are unbelievably inefficient storage-wise (I'm looking 
at an example where the file is 525K and the bzip2'd version is 29K) it 
will be great if you can handled compressed sas7bdat files too.

Thanks for the excellent work,
Frank


-- 
Frank E Harrell Jr   Professor and Chairman        School of Medicine
                      Department of Biostatistics   Vanderbilt University