Andrew James

With the way that the vanilla RNG works it's a combination of a specific `uniform random bit generator` (a `linear_congruential_engine` using the Borland C++ constants) and one of two related...

Item seed collisions can cause items to be picked up other than the target

1

**Important information** Win 11 / DevilutionX master / compiled **Describe** The way vanilla Diablo/Hellfire handle picking up items means they always do a search over the list of dropped items,...

Use WinRT to get user languages on Windows 10+

2

Turns out windows locale support was absolute garbage until recently, yay. Windows provides multiple APIs for getting user/system language settings. Of course they all seem to work differently, and the...

One possible approach for #2586, this container represents the way Items and Missiles are used (requiring both push_back and erase) as well as Monsters and Objects (typically created soon after...

Avoid segfaults when using randomly generated array indexes in items.cpp

1

The first couple of commits set up helpers for future checks. Items.cpp was one of the easier wins, drlg1.cpp is probably the worst offender so I'm trying to avoid that...

Add alternate RNG and proposed API for versioning Random Number Generators

3

This builds on the work done by @qndel in #2084. I have not updated any of the call sites identified in that PR as I wanted to make it obvious...

Read and cache robots.txt files for each host using thread-local storage

6

Would be better to share between threads to minimise requests to the same host, but that'll require synchronising read() calls and object initialisation. This also doesn't respect the request-rate or...

Matches the expectations of how noindex is treated by clients which respect robots meta directives.

https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/User-Agent#library_and_net_tool_ua_strings provides a few examples, also see urllib which uses "Python-urllib/". img2dataset does not parse HTML so has no reason to pass a user-agent that indicates mozilla compatibility.

Non-https TIO links trigger mixed-content warnings and display empty results element

2

The regex looking for the protocol/host part of TIO links (around line 773 in function breakoutTIOLinks()) picks up http and https links then performs an XMLHttpRequest with the URL using...