Tests are passing under the debugger, but not otherwise!


The "It works on my machine" joke has reached a whole new level. 😬

Down to 28 failures out of 348 tests for the URI parser.

There are other features that I haven't written tests for or implemented, so there's still work to do.

Don't pass allocators to vector<any>'s constructor.

Half a day wasted debugging.

My bugs are getting crazier and crazier by the day.

My 2000+ lines of code URI Parser is being replaced.

One of my first pieces of web++ library that I implemented.

2625 failed tests out of 4986 for the WHATWG tests I generated for my URI Parser

Is that bad?

resize_and_overwrite tip: these two values are not the same; I'm just saying to let you know before you spend an hour debugging off-by-one errors.

Down to only 2621 failed tests, yaaay 😅😬

Merged upstream changes of my liburing C++ified fork.

Still some UBs remain (liburing itself heavily relies on UB).

Header-Only liburing fork

The White House: use Memory Safe languages

Youtube Link | white house press release
Full Report (A PATH TOWARD SECURE AND MEASURABLE SOFTWARE)

White house full technical report about memory safety in languages among other things.

WHATWG mindset:

Valid URLs:

file:
any-random-scheme:

Invalid URLs:

http:
https:
ws:
wss:
ftp:

I know it's off-by-one, I just don't know in which direction it's off by one!

Rust guys said "C++ is unsafe" so much that Google's AI now thinks it's unsafe for children and refuses to help you out if you're under 18 😬🤦‍♂️🤦‍♂️

I swear, if I see one more allocator error message, ...


... I'll still spend 4 hours fixing it anyway, but I'm gonna be swearing the whole time!

If you don't know how to get it right, get it wrong a couple of times first, it narrows down the right solution, and speed things up.

Sometimes, just doing it with a few trials and errors is better than actually throwing brain cells at the problem you're trying to fix.

WHATWG tells us that http://this.is.totally.a.valid.domain.%21.