leaves | pure Go implementation of prediction part | Machine Learning library

I have two tables as follows :

Background
I have a complex nested JSON object, which I am trying to unpack into a pandas df in a very specific way.

JSON Object
this is an extract, containing randomized data of the JSON object, which shows examples of the hierarchy (inc. children) for 1x family (i.e. 'Falconer Family'), however there is 100s of them in total and this extract just has 1x family, however the full JSON object has multiple -

I recently discovered how to simulate higher order types in Java in a somewhat roundabout way like so

The function above AllPaths() appends an array containing the path to each leaf of the binary tree to the global array res.

The code works just fine, but I want to remove the global variable res and make the function return an array instead. How can I do that?

From various sources, I have come to the understanding that there are four main techniques of string formatting/interpolation in Python 3 (3.6+ for f-strings):

1. Formatting with %, which is similar to C's printf
2. The str.format() method
3. Formatted string literals/f-strings
4. Template strings from the standard library string module

My knowledge of usage mainly comes from Python String Formatting Best Practices (source A):

• str.format() was created as a better alternative to the %-style, so the latter is now obsolete
  • However, str.format() is vulnerable to attacks if user-given format strings are not properly handled
• f-strings allow str.format()-like behavior only for string literals but are shorter to write and are actually somewhat-optimized syntactic sugar for concatenation
• Template strings are safer than str.format() (demonstrated in the first source) and the other two methods (implied in the first source) when dealing with user input

I understand that the aforementioned vulnerability in str.format() comes from the method being usable on any normal strings where the delimiting braces are part of the string data itself. Malicious user input containing brace-delimited replacement fields can be supplied to the method to access environment attributes. I believe this is unlike the other ways of formatting where the programmer is the only one that can supply variables to the pre-formatted string. For example, f-strings have similar syntax to str.format() but, because f-strings are literals and the inserted values are evaluated separately through concatenation-like behavior, they are not vulnerable to the same attack (source B). Both %-formatting and Template strings also seem to only be supplied variables for substitution by the programmer; the main difference pointed out is Template's more limited functionality.

I have seen a lot of emphasis on the vulnerability of str.format() which leaves me with questions of what I should be wary of when using the other techniques. Source A describes Template strings as the safest of the above methods "due to their reduced complexity":

The more complex formatting mini-languages of the other string formatting techniques might introduce security vulnerabilities to your programs.

1. Yes, it seems like f-strings are not vulnerable in the same way str.format() is, but are there known concerns about f-string security as is implied by source A? Is the concern more like risk mitigation for unknown exploits and unintended interactions?

I am not familiar with C and I don't plan on using the clunkier %/printf-style formatting, but I have heard that C's printf had its own potential vulnerabilities. In addition, both sources A and B seem to imply a lack of security with this method. The top answer in Source B says,

String formatting may be dangerous when a format string depends on untrusted data. So, when using str.format() or %-formatting, it's important to use static format strings, or to sanitize untrusted parts before applying the formatter function.

1. Do %-style strings have known security concerns?
2. Lastly, which methods should be used and how can user input-based attacks be prevented (e.g. filtering input with regex)?
   • More specifically, are Template strings really the safer option? and Can f-strings be used just as easily and safely while granting more functionality?

I have upgraded my log4j-core dependency to 2.15.0 in order to prevent any potential Log4Shell attack. That being said I could not upgrade slf4j-log4j12's indirect log4j dependency from 1.2.17 since the latest stable version of slf4j-log4j12 is still dependent on log4j 1.2.17. This still leaves my webapp vulnerable to CVE-2019-17571 if I am not mistaken. So reading about possible mitigation strategies I came across this article which recommends to :

prevent the socket port enabled by the SocketServer class in Log4j from being opened to the public network

Could anyone please explain to me how can this be achieved and whether would this workaround be sufficient?

I'm looking for a way to have all keys / values pair of a nested object.

(For the autocomplete of MongoDB dot notation key / value type)

I've started using docker buildx to tag and push mutli-platform images to ECR. However, ECR appears to apply the tag to the parent manifest, and leaves each related manifest as untagged. ECR does appear to prevent deletion of the child manifests, but it makes managing cleanup of orphaned untagged images complicated.

Is there a way to tag these child manifests in some way?

For example, consider this push:

I use this reverse-bit method of iteration for rendering tasks in one dimension, the goal being to iterate through an array with the bits of the iterator reversed so that instead of computing an array slowly from left to right the order is spread out. I use this for instance when rendering the graph of a 1D function, because this reversed bit iteration first computes values at well-spaced intervals a representative image appears only after a very small fraction of all the values are computed.

So after only a partial rendering we already have a good idea of how the final graph will look. Now I want to apply the same principle to 2D rendering, think raytracing and such, the idea is having a good overall view of the image being rendered even from an early stage. The problem is that making the same idea work as a 2D iteration isn't trivial.

Here's how I do it in 1D: