tskit
tskit copied to clipboard
tskit-dev
Python verify_tables method?
A good number of times I have made tables that aren't valid tree sequences. When trying to convert them, I get errors such as
_tskit.LibraryError: time[parent] must be greater than time[child]
It would be really helpful to know which nodes / edges were causing the problem (it's easy in this case, as I can run through the edges, but for some other tree sequence properties it's less obvious). I assume we have python code in the test suite that can help to debug this. I wonder if a separate (slow) routine, tables.validate() or similar, which could output more specific error messages, would be helpful here?
I agree, it would be very helpful but is a bit of a slog to actually do.
Also, do we have an issue for this already?
Also, do we have an issue for this already?
Hmm, I didn't find one, but didn't look hard either.
No, I can't find one either. Must be something we discussed in one of the long threads.
What we probably want is something like assert_tables_equal where we first call the C function to make the quick check, and then if that fails we slog through the various ways it could fail.
So, Tables.assert_valid() or something?
Great idea, would be quite a nice thing for a contributor to do as it is pure python and the C routines can be used a reference. I don't see me getting to this anytime soon.
