darshan-hpc / darshan

Darshan I/O characterization tool
Other
56 stars 28 forks source link

MAINT: "data access by category" algorithms/data structures vs. plotting separation #681

Open tylerjereddy opened 2 years ago

tylerjereddy commented 2 years ago

From a comment I made in gh-397:

it would be nice to shift some of the sorting and numerical manipulation outside of the plotting functions and into more "algorithm library" type functions, possibly by organizing the latter into a separate module

Someone might have to be pretty motivated to tackle this relative to other higher priorities though.

tylerjereddy commented 2 years ago

Part of the longer-term motivation might be preventing bloat in the test suite--for example, try to minimize slow plotting tests that place pixels into an object, when the basic thing they are testing could be captured by moving the data munging to a numerical module/function that is separable from plotting infrastruture/matplotlib.

We'd still want plotting tests and regression tests when plotting-specific issues crop up, but the way I've organized it probably leans more heavily on producing matplotlib figures and then retrieving numerical/text data from those objects, instead of loading most of the testing at the "source"/data structure side.