Re: [ww-users] R off-wiki

[Jonathan D. <du...@mc...>]

I'm not sure how this relates to your vision, but I do the part
described at the beginning all the time.  I download a .RData file
from the wiki and play with it until I have an idea what I want to
code on the wiki.  It's fairly straightforward.  Just save an .RData
file to my local disk and say 'R --vanilla' and then
'load("wiki.RData")'.  It's simple enough that I don't feel very
tempted to worry about the rest of it in the short term.  In the long
term, we might aim for a broader approach to treating WW files as
files

I'll mention too that R can already read both .R and .RData files
directly from publicly readable wikis, so the only possibly
non-trivial obstacle to the vision below is passwords.

Another trick that I guess is worth sharing is that I rarely click on
a .make.log file when debugging.  At the first sign of trouble, I put
temporary tags for "target.Rout" and "target.Rout.make.log" (and often
"target.dmp") right at the top of the page.  It makes a surprisingly
large difference in the "feel" of the interaction.

JD

On Mon, May 23, 2011 at 11:53 AM, Lee Worden <wor...@gm...> wrote:
> Imagine you are working on R programs in a project in a wiki.  You start up
> R on your own workstation, load in the data and functions from the files on
> the wiki, and try out R commands until you're satisfied, then go back to the
> wiki and update it.  Much easier than editing, saving to the web browser,
> previewing, reloading the .make.log file, right?  Also easier than exporting
> the whole project, unpacking the tar file, and loading from there.  It may
> be doable without any changes to WorkingWiki.

> I think it would just require writing an R library that can load data files,
> .RData files, and maybe .R files from wiki locations into your R session.
> It might have functions like attachToWWProject(wikiname, projectname),
> loadFromWW(rdatafilename).

> If anyone feels like writing something like that, I'd be happy to consult
> with them.  I don't have experience writing R libraries, and I won't have
> time to do something like this anytime soon.