[Biojava-dev] Closing ResultSets and Statements in biosql package.

Keith James kdj at sanger.ac.uk
Fri Aug 1 16:15:20 EDT 2003


>>>>> "Matthew" == Matthew Pocock <matthew_pocock at yahoo.co.uk> writes:

    Matthew> Len Trigg wrote:
    >> I don't see why a test couldn't just repeat some large number
    >> of times on the same sequence, but it might be a bit of a slow
    >> test.  Cheers, Len.
    >> 

    Matthew> As long as the unit tests for db access are not run on
    Matthew> 'ant runtests', I don't think it matters how long they
    Matthew> take to run - the important thing would be to ensure that
    Matthew> the database access is robust and produces the right
    Matthew> results - particularly since we would be putting large
    Matthew> amounts of important data inside biosql, or another one
    Matthew> of these db-backed sequence dbs.

For sure. We could have an 'ant rundbtests' as a target which has to
be run explicitly.

    Matthew> Is it possible to re-use the top-level test cases for
    Matthew> biosql, ensembl, and any other biojava SequenceDB
    Matthew> implementations we want to test to destruction?

It should be possible to do this, plus having separate targets for
each platform ('ant runbiosqltests', 'ant runwhatevertests') under the
'ant rundbtests' umbrella. Then you could run a specific subset of the
dbtests, or run the whole lot.

Maybe it's even time to have a tests.xml file to stop the build.xml
bloating further - I think that the core tests should stay in
build.xml, but specialist/optional ones could go in a new file?

Keith

-- 

- Keith James <kdj at sanger.ac.uk> bioinformatics programming support -
- Pathogen Sequencing Unit, The Wellcome Trust Sanger Institute, UK -



More information about the biojava-dev mailing list