Hello,
It is possible this is somewhere in the documentation, but I could not find it. I have a recursive strategy, and I would like to generate very large recursive structures with a low amount of examples. However there don't seem to be a way to force hypothesis to start generating large values.
Another framework I used has a max_size parameter, where sizes go from 0 to max_size over the number of examples, so if we want faster growth that can be set to some larger number as needed.
There is a couple of options here, either we could allow for specifying some size parameter in the environment (like @settings for example), or there could be some strategies for scaling other strategies like scale in this other framework. Lastly there could be a min_leaves added to recursive.
Also, just to be clear, I much prefer hypothesis over the other framework, it is only used as a reference because it allows for manipulating the underlying size concept more directly, which is something that would simplify my use-case a good bit.
My use-case is essentially that there is a significant overhead to running a single example, while the impact of a large input is insignificant by comparison.
EDIT: also this is a question/enhancement, could not figure out how to set labels
Hello,
It is possible this is somewhere in the documentation, but I could not find it. I have a recursive strategy, and I would like to generate very large recursive structures with a low amount of examples. However there don't seem to be a way to force hypothesis to start generating large values.
Another framework I used has a
max_sizeparameter, where sizes go from 0 tomax_sizeover the number of examples, so if we want faster growth that can be set to some larger number as needed.There is a couple of options here, either we could allow for specifying some size parameter in the environment (like
@settingsfor example), or there could be some strategies for scaling other strategies like scale in this other framework. Lastly there could be amin_leavesadded torecursive.Also, just to be clear, I much prefer hypothesis over the other framework, it is only used as a reference because it allows for manipulating the underlying size concept more directly, which is something that would simplify my use-case a good bit.
My use-case is essentially that there is a significant overhead to running a single example, while the impact of a large input is insignificant by comparison.
EDIT: also this is a question/enhancement, could not figure out how to set labels