[Phylobase-devl] anyone there?

Jim Regetz regetz at nceas.ucsb.edu
Thu Aug 20 20:06:08 CEST 2009


François Michonneau wrote:
> On Thu, 2009-08-20 at 12:30 -0400, Ben Bolker wrote:
>> 6. testing (#589): finish moving tests to RUnit framework??  what is the
>>     status?  I need someone to explain a little more to me about how
>>     this works, I have questions ... is a complete move to RUnit
>> necessary for release?
> 
> It seems to me that Jim pushed most of the changes last night. He moved
> my tests to the correct directory.

Yes, I did move François' pre-existing RUnit tests. But I haven't 
migrated any of the ad hoc tests in the tests/ directory into the RUnit 
test framework. Or, I've started to, but haven't committed anything.

Current issues:
1. It's hard to figure out what is tested, and where
2. Tests aren't terribly thorough yet
3. Some things are now tested in multiple places

What I'd like to complete before the release:
1. Restructure the unitTests directory to mirror the *.R source files in 
the pkg/R directory. Each of these test files will, in turn, have a test 
function for each method/function in the corresponding .R source file. 
Those test functions will be trivial in some cases, but more extensive 
in others (testing edge cases, different argument combinations, etc).

So for example, pkg/inst/unitTests/runit.methods-phylo4.R will 
correspond directly to pkg/R/methods-phylo4.R. The test file will 
contain a test function for each method/function in the source file. 
I'll commit a skeletal revision of this test file in a minute, to 
illustrate.

2. Move many tests out of the existing ad hoc pkg/tests/*.R test files, 
and into the RUnit tests.

After this re-org, I hope it will be clear where you should go to 
add/update test cases if you add/modify code in some way.

Then I expect there will be other tests that are more of the integration 
test variety rather than unit tests. There is no reason these couldn't 
also be written up in the RUnit framework, but we could also leave them 
as tests/*.R files. We certainly don't need to figure this out before 
the CRAN release, as far as I'm concerned.

Cheers,
Jim


More information about the Phylobase-devl mailing list