[Biopython-dev] Jython 2.7b2 - test_NCBITextParser.py and test_Pathway.py

[Peter Cock]

On Wed, May 7, 2014 at 5:50 PM, Peter Cock <p.j.a.cock at googlemail.com> wrote:
> Jython lives! http://fwierzbicki.blogspot.co.uk/2014/05/jython-27-beta2-released.html
>
> I just tried Jython 2.7b2 on Mac OS X, and mostly things work :)
>
> test_NCBITextParser.py fails due to some long methods, see
> http://bugs.jython.org/issue1891 and our past hacks, e.g.
> https://github.com/biopython/biopython/commit/deb14b29ac35b9ed74911bf0c17355a0e9bc4f21
> https://github.com/biopython/biopython/commit/7e614c149a1bf0dfe80d77b1021a6cca508fdd10
>
> (This looks like a regression in Jython being stricter here, since
> the error doesn't say which method(s) are too long, fixing this
> is a bit painful... perhaps counting lines per method would
> spot the problem methods nicely?)

Does anyone fancy that fairly boring task? We'd like to get
rid of the NCBI plain text BLAST parser at some point - but
anecdotally people continue to use it.

> Separately, test_Pathway.py is failing which looks more
> interesting... details below.

This appears to be due to our (old) code using "is" and a
possible regression in Jython: http://bugs.jython.org/issue2137
https://github.com/biopython/biopython/commit/2d1e14722907b82dae1afedabad72791019958ca

I think we should be using "==" rather than "is" here anyway,
but there might be some downsides with non-string graph keys?

Peter