FastAPI, Concurrency, and Python Performance Hacks

Key Quotes

"Normally I don't mention this but both of these vulnerabilities have CVE numbers CVE is short for Common Vulnerability Exposure these are public IDs given to a security issue which helps researchers coordinate and track problems in the field for example the http response redirect vulnerability that I just mentioned is officially CVE 202564455644558 or Bob to its friends"

Christopher Bailey explains that CVE numbers are public identifiers for security issues, facilitating coordination and tracking among researchers. He notes that these numbers are assigned by specific authorities, and highlights that Django has now become a CVE Numbering Authority (CNA) itself.

"Last year the Python Software Foundation board of directors started holding office hours once a month an opportunity for you to talk to them sometimes they cover topics and other times they just take questions but if you want to know more about the PSF and the board and how things are going you can attend one of these sessions"

Christopher Bailey informs listeners about the Python Software Foundation (PSF) board's monthly office hours, presenting them as an opportunity for community members to engage directly with the board. He suggests attending these sessions to gain insight into the PSF's operations and current status.

"In this particular case you'll be creating a virtual environment and installing FastAPI into it and this is actually one of the first of many intricacies for the tutorial when you go to pip install you need to specify what you want in your particular install of FastAPI the framework itself in this case but you also want to have this thing called an ASGI ASGI asynchronous server gateway interface and in this case it's using a package called uvicorn which can be included you just have to specify that"

Christopher Trudeau details a step in building a FastAPI application, explaining that installing FastAPI requires specifying not only the framework but also an ASGI server like uvicorn. He notes this as an early intricacy in the process, highlighting the need to explicitly include these components during installation.

"There are two kinds of concurrency IO bound and CPU bound let's talk about IO bound for a bit your processor spends a lot of time waiting for things it's actually kind of hard to wrap your head around the scale of the difference to try to hit that home in the course I cover a distance analogy so instead of time think of distance let's equate a single CPU instruction that's about 0.001 nanoseconds depending on your processor etc etc your mileage may vary let's call that single CPU instruction about a meter"

Christopher Trudeau introduces the concept of concurrency by distinguishing between I/O-bound and CPU-bound tasks. He uses a distance analogy, equating a single CPU instruction to a meter, to illustrate the vast difference in scale when processors spend time waiting for operations like memory references or network requests.

"When you make an instance of a class it has attributes I thought of an example like let's say you have products for an online store let's say if you sell t-shirts you create an instance of the shirt class and it has attributes for the t-shirt the design the price and the size and so you'd normally access them with dot notation so you'd be like shirt.design and equals this and shirt.price equals and shirt.size equals you know whatever the letter is for the size"

Christopher Trudeau explains that class instances in Python store attributes, using the example of a shirt class with attributes like design, price, and size. He clarifies that these attributes are typically accessed using dot notation, such as shirt.design.

"The first pointer is about knowing when to use sets in Python if you're doing a lot of membership lookup lists are actually fairly expensive and you'd be surprised how often the order of things really isn't important in your code and if the order isn't important you can get away with using a set and that might boost your lookup performance"

Christopher Trudeau highlights a performance hack for Python code, advising the use of sets for membership lookups when the order of elements is not critical. He explains that lists can be expensive for such operations, and switching to sets can significantly boost lookup performance.