_gibbs fails on dataframes with floats

[lkursell]

_gibbs fails with:
IndexError: index 6283 is out of bounds for axis 0 with size 6283
When the are floats in the sink_df. The error occurs when trying to create the order to walk through the sequences.

Probably want to add a check so that a less cryptic error message is given to the user.

[wdwvt1]

Can you post the dataset + command that causes this. I thought I had every error where an index error occurs isolated, but definitely need to see what slipped through.

[wdwvt1]

this can be resolved by changing line 529 in sourcetracker2/sourcetracker.py to be sink_sum = np.int(sink.sum()). alternatively we can floor or ceiling this value, but we just need to ensure its not non-integer.

[lkursell]

Should we make the user address their int problem? or just call floor and add a note saying that is what we do?

On Apr 26, 2016, at 3:52 PM, Will Van Treuren [email protected] wrote:

this can be resolved by changing line 529 in sourcetracker2/sourcetracker.py to be sink_sum = np.int(sink.sum()). alternatively we can floor or ceiling this value, but we just need to ensure its not non-integer.

—
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub #41 (comment)

[wdwvt1]

This is now resolved in PR #51 and noted in the readme.md.