Tests necessary for 1.4

lmaurits / BEASTling

A linguistics-focussed command line tool for generating BEAST XML files.

BSD 2-Clause "Simplified" License

20 stars 6 forks source link

Tests necessary for 1.4 #202

Open Anaphory opened 6 years ago

Anaphory commented 6 years ago

[ ] self.distribution.dist == "point"; bad value for parse_prior_string and point distributions in distributions.py
[ ] self.config.path_sampling
[ ] self.birthrate_estimate is not None
[x] covarion model with bad value for share_params and share_params==False

Bedhead015 commented 6 years ago

Hello, I'm a student and I would like to work on this issue, for my software testing class.

lmaurits commented 6 years ago

Okay, this might be possible! Some of the tests probably require a little bit of understanding of the modelling that BEASTling is doing (do you have any kind of interest/background in statistical modelling and/or phylogenetics?), but others are pretty straightforward and just about making sure that certain values for certain parameters do or do not throw an exception. Have you managed to clone the BEASTling repository and run the existing test suite?

@Anaphory, I assume these tests are actually still needed and we didn't just forget to close this issue when we cut 1.4.0 earlier in the year?

Anaphory commented 5 years ago

@Bedhead015 Welcome!

If there are particular things you want/need to do for you class, we can surely find them here or elsewhere in the project. If you want to discuss what you need and what you would like to do, both I and @lmaurits have email adresses on our GitHub profile.

I'm very sure that these are my extremely terse notes/reminders for the tests still needed. If I had worked on them, I would have done that with enough dedication to actually tick the boxes. Point in case: there are no file names like bad_share_params.conf in https://github.com/lmaurits/BEASTling/tree/master/tests/configs/bad_configs.

Given that these are to some extent just personal reminders, it may take a moment to figure out what they mean, but definitely the last one, and to some extent the other ones, are mostly writing configuration files with those settings and adding them to the list of expectedly bad or good configurations.

Anaphory commented 5 years ago

On the topic of tiny tests: https://github.com/lmaurits/BEASTling/blob/b422d1644e118d87e78c50e7d50194bb21f69edb/beastling/beastxml.py#L208 does not have an else clause for strange values for tree_prior. I would like to see a test that an invalid tree prior name fails with a clean error message.

[x] tree_prior = no_such_tree_prior This may be a new change in #208 which is an entirely different piece of cake, though.

Anaphory commented 5 years ago

[x] I just fiddled with tree priors a bit more, and I'm pretty sure that https://github.com/lmaurits/BEASTling/blob/b422d1644e118d87e78c50e7d50194bb21f69edb/beastling/beastxml.py#L591 is not visited by our tests, though I have not looked at its coverage report. (I moved it somewhere else, where tracer_logger was undefined, and did not get any complaints.)

Anaphory commented 5 years ago

The Yule birthRate logger is definitely visited now: https://codecov.io/gh/lmaurits/BEASTling/src/tree_module/beastling/treepriors/base.py#L214