This is an archival dump of old wiki content --- see for current material

Working on Scipy

Anne Archibald's workflow

This is on Linux.

The first time

Check out the current SVN version

svn co to some local directory

Set things up so compilation doesn't break anything

write to do a setuptools install to some arcane local directory; write to source when you want that directory on your PYTHONPATH

Compile the current version


Run the tests

python -c "import scipy; scipy.test()"

Unfortunately, there are often some failing tests before you even start. Usually I just check that nothing looks too suspicious and remember that these ones are still going to fail when I'm done.

Second and later times

Update SVN

svn update

Compile and test

python -c "import scipy; scipy.test()"

Making changes

Write some tests

How to test the module you're working on, say scipy.interpolate:

./ && python -c "import scipy.interpolate; scipy.interpolate.test()"

How to write a test:

How to make sure it's getting run: I usually just add "assert False" and make sure it fails.

How to run just one test and bring up a debugger if it fails:

Fix the bug/implement the feature

This is usually surprisingly easy by this point.

Run all scipy tests

python -c "import scipy; scipy.test()"

Document the change

Just write some docstrings. I usually just format them like other docstrings in similar functions, but there's a prescribed format somewhere.


svn commit

Force an autobuild:

SciPy: WorkingOnScipy (last edited 2015-10-24 17:48:26 by anonymous)