Pythonista | Collection of Python Scripts written for Pythonista iOS App

Probably went through else statement, that does not return anything. Tried your code and it works as expected if you meet if criteria in bet function (one that has return)

j <= lengthData will let j reach a value beyond the last index in data which is lengthData-1 because Python list indexes are zero based.

Use while j < lengthData: instead

Oops, you are playing with implementation details here. What the language says is that an object as a unique id, and if two objects have same id, then they are the same. This is mandated by the language.

Whether the id of a deleted object will be re-used by the next created object is clearly a detail, depending on the implementation and possibly on other details. So it may be the same or not, and without digging into the source of an implementation you cannot be sure of the result.

Only one thing is sure here, you should not rely on getting the same id.

You are missing an opening bracket on the database name. It should ideally be like this:

For your problem you want all your (pointwise) errors to be close to zero. So... just use allclose on the error timeseries and a zero vector (it broadcasts under the hood):

Where does the dataframe come from - how was it generated? Was it imported from a CSV file?

Your code works if you apply it to columns which are strings, as long as you remember to do df = df.apply() and not just df.apply() , e.g.:

Pythonista does not support pip installing binaries and has no compiler for C code.

There is a pure Python alternative pg8000 which at least 1 Pythonista user has reported to work.

You say IDLE, which Pythonista does not have, so I am assuming that you are trying this outside Pythonista. ui module is a proprietary part of Pythonista, and not installable separately.

If you want to use UIKit classes with Python on a Mac, you either need a new M1 Macbook that can run the same Pythonista app that you run on your iPad, or you have to look at something like Rubicon to bridge from Python to the UIKit classes, essentially duplicating the Pythonista ui module functionality.

If pbkdf2 is natively available, I'd use that before trying to roll your own bcrypt. When its work factors are sufficiently large, it's still a solid choice when bcrypt or scrypt aren't available, and using it directly is safer than trying to recreate something else by hand.

Not knowing more about your use case, a general recommendation: use pbkdf2 with a sufficiently large number of rounds to take about a half-second's worth of the upper end of the processor throughput of your target devices. This keeps the UX within tolerable wait times while still providing reasonable resistance to offline attack.

I'd also recommend randomizing that number of rounds slightly over a range (like a thousand). For example, if you settled on 200,000 as having an acceptable 500ms delay, I'd randomly pick a value between 200,000 to 202,000 (or something like that) - whatever is needed to ensure that most users will have different rounds from each other (assuming that all user passwords might be aggregated into a single location that could be compromised and the hashes stolen). This is because some of the newer "associative" / "correlation" attacks only work well against a large set of hashes when all of the cost factors across that set of hashes are the same.

Long term, also be sure that your code easily accepts a variable floor and ceiling for the number of rounds, so you can choose to increase your number of rounds over time as processors advance. (You could even get fancy and dynamically calculate the range of rounds based on the processor that the password is being created on, so that it's future ready without any additional intervention.)