PooCon2016: Testing and debugging Pootle

testing with pootle

• coverage is up to around 93%
• py.test
• high test coverage facilitates rapid dev

CI tools

• travis.com
• coveralls.io
• codecov.io
• landscape.io
• requires.io
• New Relic

introduction to pdb

• swiss army knife of debugging
• introspect local vars and test code
• import pdb; pdb.set_trace()
• import inspect; inspect.stack()[10]

sql debugging

• usually the most important part of optimization
• debug lines of django code with debug_sql
• use the sql in dbshell and experiment
  • describe select ...
  • show create table ...
  • alter table ... add index ...

debugging tests and optimizing

• py.test -vv --debug.tests my-timings.txt
• drill down to time-consuming tests and debug_sql
• common optimizations
  • preventing duplicate calls
  • adding select_related or pre-initializing related objects
  • prefetching
  • values_list - sometimes with a ProxyAdapter (also Queryset.only() ?)