Strange behavior with bayesfactor_ttest for weak effects with one-sided test

[dsuplica]

The documentation for bayesfactor_ttest states that

One-sided Bayes Factor (BF) are simply obtained by doubling the two-sided BF, which is not the same behavior as R or JASP. Be extra careful when interpretating one-sided BF, and if you can, always double-check your results."

While this is generally correct, it does not completely follow for t values translating to bayes factors below BF = 1. Essentially, for a two-sided test which corresponds to a BF < 0.5, the BF for alternative="greater" and alternative="less" will be identical. I have included a curve of the outputs of bayesfactor_ttest corresponding to different alternative hypotheses:

Essentially, within a certain range of weak effects, the one-sided BF appears to be uninterpretable as a result. I believe it is due to this code in bayesian.py, where it will only invert the BF if it is greater than 1

tail_binary = "two-sided" if alternative == "two-sided" else "one-sided"

bf10 = bf10 * (1 / 0.5) if tail_binary == "one-sided" else bf10

# Now check the direction of the test

if ((alternative == "greater" and t < 0) or (alternative == "less" and t > 0)) and bf10 > 1:

        bf10 = 1 / bf10

return bf10

I am not familiar enough with bayesian statistics to necessarily know what the "correct" output should be. However, I would have at least expected to see a monotonic curve for the one-sided BF as the t value changes. This makes it quite hard to see evidence in favor of a null when there is an a priori prediction about the data. Should I be only using a two-sided test, even where this would be the case? If that's true, it might be good to have an explicit warning about this beahavior.