Hi there, Baiju, much thanks for looking into this. I hope we can indeed get rid of this code. I myself have the suspicion that the deprecation system is perhaps a 'false optimum' in most cases. Putting in deprecations tends to be quite a bit of work (as it's a code change), the warnings weren't always very instructive in the past, and we haven't done a good job of removing deprecations over time. Instead it might be better if we use that energy to provide better documentation about changes and what to do in plain text. Code might break where the deprecation system would provide a backwards compatibility warning, but perhaps that's a faster and easier way to get people to update their code. For deprecated import locations I recommend we just put in a backwards compatibility import (no deprecation system) and use a combination of documentation and the enhanced test runner which can report about indirect imports. Perhaps there are other cases where the deprecation system is of use, but I myself am quite inclined to see about ripping it out and see what happens. Tres Seaver wrote: [snip]
The 'compattest' suff which the sprinters added last month would be a place to start: it runs the tests for dependency packages using your locally modified package.
For more information see here: http://pypi.python.org/pypi/z3c.recipe.compattest If you can't get it to work right for you (it's tricky business to get it set up), or have any other suggestions for improving it, please start a thread here. I'll get the right people to listen. Regards, Martijn