[Phylobase-devl] Aug. virtual hackathon

Peter D. Cowan pdc at berkeley.edu
Thu Aug 13 21:36:54 CEST 2009


On Thu, Aug 13, 2009 at 10:23:10AM -0700, Jim Regetz wrote:
> Peter Cowan wrote:
> >Before, we actually start coding, it's probably worthwhile having
> >at  least some discussion about the changes that Francois has
> >made, and  any other major changes people would like.  It would
> >also be good to  get the RUnit stuff checked in (if I can help at
> >all with that let me  know Jim) so that we can make a good dent in
> >writing unit tests (which  will make any future changes much
> >easier to carry out).
> 
> Sure thing, I'll go ahead and check in what I've got. Sorry for the
> delay. I see François has committed some RUnit tests in his branch
> (fantastic!), so I think merging (and augmenting) our tests will in
> turn help ease merging his branch and making subsequent changes.
> That said, I should point out that I really only wrote a small
> smattering of simple tests. More will be needed. And of course if
> someone feels strongly about reorganizing the tests or changing
> their level of granularity, I'm open to it.

Great!  Merging shouldn't be too difficult (hopefully).  One of the reasons that I wanted to have it committed before we start the hackathon, is so that all of us could help with writing additional tests.  Eventually I'd like to get to the point where new features, and bug fixes get committed with appropriate unit tests.

> My commit will also include a script that should automatically
> trigger the RUnit test suite when doing a package check, plus a
> Makefile that makes it easy to run them from the command line
> whenever you want. These both come straight off the R-project wiki:
> 
> http://wiki.r-project.org/rwiki/doku.php?id=developers:runit

Also excellent, this was the guide I was using for RUnit also.  I agree that it's a good idea to run all the unit tests when R check is run.

> I've used something slightly more customized before, but I figure we
> should start with this and then tweak if needed. Works for me
> locally, but after I commit it we'll probably just have to wait a
> day to see if breaks the RForge checks.

I've also hacked up a script to run R CMD check on the package every time there is a commit to the trunk.  It's not completely functional yet, but before Tuesday I plan to set  it up on my Lab's sever (which unfortunately takes a full 5 min to run check).

Peter


More information about the Phylobase-devl mailing list