Remarks on dumpdata() and data.dump()

From lab 1 you already know how to use dump to save copies of objects for later use in Splus. For example, if you want to save a data.frame called fred and a related function called make.fred you could type:

dump(c("fred","make.fred"), file = "A:\\myfile")

Provided you had a floppy disk in the A: drive, you would then have a file called myfile on that disk. It would be a text file; you could look at it with an editor.

You could give your disk to a friend, who might copy the file to something like "C:\\yourfile" (no need to use the same name), or to some other place where your friend is allowed to write. From within Splus, your friend could then type

source("C:\\yourfile")

to have Splus recreate both the fred and make.fred objects in the working directory. Your friend could also work directly from the floppy:

source("A\\myfile")

If some kind person puts a copy of a file in dumpdata format on the web, you have only to save it to a convenient place as a text file, then source() that textfile to get the Splus objects in your own working directory.

Unfortunately, dump() and source() can be rather slow. I have even managed to crash a StatLab machine by trying to source a huge dumpdata file. The solution to the problems of dump/source is to work with the close cousins:

data.dump(c("fred","make.fred"), file = "A:\\myfile")

and

data.restore("A:\\myfile")