[R] open source and R

[Martin Maechler]

>>>>> "Jari" == Jari Oksanen <jarioksa at sun3.oulu.fi>
>>>>>     on Tue, 15 Nov 2005 12:25:17 +0200 writes:

    Jari> On Tue, 2005-11-15 at 09:54 +0000, Barry Rowlingson wrote:
    >> Liaw, Andy wrote:

    >>> However code readability can not be over-emphasized.  I
    >>> must admit to have written R code in such a supposedly
    >>> `clever' way that I can't figure out what I was trying
    >>> to do (or how I did it) a week later...

    Baz> The solution to that is to make sure this sort of code
    Baz> is adequately commented! Be as clever as you like -
    Baz> make your R look like a runner-up in the obfuscated
    Baz> perl programming contest if you want - but a
    Baz> well-placed comment will hopefully prevent that stupid
    Baz> feeling a week later.

Exactly!

    Jari> Unfortunately the comments don't stick well with the R
    Jari> code. They would if you always edit the source code,
    Jari> but not with my preferred toolbox.  For me the most
    Jari> natural way to work on a function is to install the
    Jari> package with the function, and then use Emacs+ESS to
    Jari> edit, test and debug the the function within an R
    Jari> session. 

But why?  Why on earth are you not working with the *.R files in your
    <pkg>/R/ directory?
Or make a copy of these and work with the copy?

Also, you can use  library(<....>,  keep.source = TRUE)
and this helps for all those packages that did *not* use
lazy-loading or saved images at their installation time;
unfortunately, that excludes many packages.

Once you've seen the light, i.e. ESS :-) ,
using edit() , fix() and all those abominations is just plainly
wrong in my (biased) view!

    Jari> session. That really ruins all decent commenting:
    Jari> comments may be misplaced, and the default formatting
    Jari> of comments is really bad in ESS. 

Huh??  I'd have expected you to say the contrary here.
I assume you have never heard of the difference between "#", "##"
and "###" in comments.  Most of the R core developers adhere to
it which you can see when browsing the R source files.

Since one version of the ESS manual is online, look here:
http://ESS.R-project.org/Manual/ess.html#index-comments-in-S-154

    Jari> So my choice is to uncomment R code, but comment C
    Jari> (and Fortran).

ooh;  I hope you can be convinced to use comments in R code...
and maybe adapt to the "##" vs "#" (or "###") scheme.

Regards,
Martin Maechler, ETH Zurich