[Biopython-dev] Python 3 status (ignoring numpy and our C code)

[Peter]

Hi all,

I haven't gotten round to installing NumPy under Python 3 on this machine.
Summary of test output (ignoring all the passes and skipped tests) using
2to3 with default settings.

------------------------------------------------------------------------

test_CAPS ... ERROR
test_Restriction ... ERROR

TypeError: unhashable type: 'RestrictionType'

This is a tricky issue, see:
http://lists.open-bio.org/pipermail/biopython-dev/2010-July/007975.html

------------------------------------------------------------------------

test_Crystal ... FAIL

Slicing issues, we could fix them or just deprecate Bio.Crystal
http://lists.open-bio.org/pipermail/biopython-dev/2010-July/thread.html

------------------------------------------------------------------------

test_LocationParser ... Syntax error at or near `467' token

Something in the spark parser isn't handled by 2to3, not urgent as I want
to deprecate Bio.GenBank.LocationParser which is the only thing using spark.

------------------------------------------------------------------------

test_NCBI_BLAST_tools ... FAIL

Not Python 3 specific, the latest BLAST+ has changed some switches.

------------------------------------------------------------------------

test_PhyloXML ... FAIL

Longitude versus long problem with 2to3:
http://lists.open-bio.org/pipermail/biopython-dev/2010-July/008071.html

------------------------------------------------------------------------

test_SeqIO_index ... ok

Test passes but is very very slow, see:
http://lists.open-bio.org/pipermail/biopython-dev/2010-July/008004.html

------------------------------------------------------------------------

test_SeqIO_online ... FAIL

May need to turn all online byte handles into unicode handles,
http://lists.open-bio.org/pipermail/biopython-dev/2010-July/008076.html

------------------------------------------------------------------------

test_property_manager ... FAIL

I think this is a change to the default object's __repr__ method,
and/or module name vs __main__ but in any case I'm tempted
to deprecate Bio.PropertyManager because we don't really use
it and I don't understand it ("Here be dragons!")

------------------------------------------------------------------------

Not looking too bad. Now I really should install NumPy on this machine...

Peter