[rkward-devel] rkwardtests package: done :-)

[Thomas Friedrichsmeier]

Hi,

On Thursday 07 October 2010, meik michalke wrote:
> the recent SVN version of the rkwardtests package now works.
> rktest.makeplugintests() now has a new "testroot" element to specify the
> path to the test suite, as well as the wished for "append" and "test.id"
> elements.

cool. Thanks!

> speaking of potentially malicious code: i noticed that
> rktest.cleanRKTestSuite() (see internal functions) in it's current state
> can be quite dangerous. it deletes every file that is not a test standard
> file inside directories called getwd()/suite at id. fortunately, that caused
> it to stop before it would erase my $HOME when i tried
> rktest.makeplugintests() with an absolute path to the test suite file,
> before i made testroot mandatory. this should probably be revised to a
> more robust workflow, but with the added need for a testroot definition
> it's not a top priority issue at the moment.

True, that should be made more robust, somehow.

> > Also, with version 0.1-2 of roxygen, I can simply write
> > 
> >   [in directory rpackages]
> >   R CMD roxygen rkwardtests/
> 
> doesn't work for me, unfortunately -- it stops with a "command roxygen not
> found" error. i installed the package to local libs and they're ignored,
> although R_LIBS and R_LIBS_USER are set accordingly (when i start R itself,
> all packages are found). i figured i could use my time more productively if
> i just skipped the effort and stick with roxygenize() ;-) but if anyone
> has a hint how it could work from command line, i'm open to suggestions.

Hm, do you have the file
  /usr/lib/R/bin/roxygen
? Possibly a problem with multiple installations of R? Check wether
  which R
is the one you expect (this one keeps tricking me again and again...).

> >   R CMD roxygen rkwardtests/ rkwardtests/
> > 
> > to make it write the new .Rd files in place).
> 
> i actually avoid that on purpose to be able to diff pre and post change
> docs, if i desire to. but that's just me ;-)

Well,
  svn diff
does that for me ;-). Of course only once you've committed a first version.
 
> > I think that could shorten your script, considerably.
> 
> hm, what do you mean? it would surely replace two "cp *.Rd" lines with one
> "R CMD" line (and shorten the R function to the system call), but not
> affect the remaining parts, would it?

Ok, I did not read your script right. I thought the "packing" was part of 
"roxygenizing", but it's not. Anyway, the "roxygenizing" becomes a one-liner 
this way - provided 'R CMD roxygen' works... I think I'll add that to 
makedist.sh for good measure, in case we ever forget to roxygenize after an 
update.

Skimming through the diff, I stumbled across this:

 # counterpart to rktest.initializeEnvironment. Restores the most important 
settings
 rktest.resetEnvironment <- function () {
-       rm (list=c ("rk.set.output.html.file", "rk.get.tempfile.name", 
".rk.make.hr"), envir=globalenv ())
        .rk.rerun.plugin.link <<- .rk.rerun.plugin.link.replacement
 }

Was that intentional?

Regards
Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/rkward-devel/attachments/20101007/94019526/attachment.sig>