Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
mrthreshold behaviour with exotic uses #1481
Currently the input to the
There's also the presence of the
I can also see other areas of code where the behaviour in the presence of non-finite input data may not be as expected.
Given this, the discussion in #205, and the bug report in #738, I think the structure & interface of
Yep, I think a review is in order. This command is ancient and has seen a few changes over the years, so a lot of the structure and interface would be relic-status. Worth a revisit - when we find the time... I guess the best thing to do at this stage is to come up with a replacement for the command-line help page, and go from there...?
My concern looking at the code was that it was going to in fact be quite difficult to describe completely accurately exactly what the behaviour of each command-line option was going to be, due to the confounds listed above behaving differently for each command-line option, and therefore describing what we want the behaviour to be might in fact be easier than trying to describe what's there already.
My initial instinct was to stick with a single threshold with the option to invert the output, rather than the "above/below" approach described in #205, with more complex masking (e.g. values within a range) achievable using
In that case, the remaining issues are:
If the above/below approach is instead used:
Other points to consider (which apply to both possibilities):