I love that someone is actively working on components in the python space. Coming back to python after years in the js ecosystem I've found templates really cumbersome to work with, especially not having type system support. I know there have been some attempts at JSX in python over the years but those projects all seem dead.
My advice to the OP, is to break out the component piece of the framework into a standalone lib that other frameworks can use. I think it'll be much easier to gain adoption that way.
I've found it's autocomplete really bad. It often hallucinates import paths and other illegal operations. It's not clear to me why it can't leverage the LSP better.
I just moved to the USA from Europe and have found sms/mms to be annoying. I'm consistently missing texts especially mms. Maybe my settings are slightly off or the payload is exceeding MMS size limits but it's way less reliable than Whatsapp. Everyone seems to use iPhones so they don't hit this issue until they text an android user like me.
I've been building up a vintage Fuji and have been using Sheldon's site to find data on old standards. It's been incredibly useful. On a similar note I've also used https://www.disraeligears.co.uk/site/home.html to look up derailleur specifics there is a lot of info there.
I had a quick look at the code to understand what (sort of) meant. It's appending to a file and then reading it back and looking for the key line by line so it's O(n). I'm not sure about redis internals but I imagine its using some search tree for saving to disk which will likely be O(log n). So in a benchmark with more data redis should win by exponential margins. Ofc if you are only storing 100s of keys then this is a good approach.
I think you are misinterpreting the term "diff". Near the beginning of the article they describe it "At Meta we call an individual set of changes made to the codebase a “diff.” So diff == PR.
I've been building a basic sqlite like database [1] to improve my understanding of relational dbs. It's still early days but it's been fun using different data structures than I typically do in my day job.
Wasn't tangram previously an easy to use ML tool? I thought the product looked pretty neat, would be interested in reading a post on the pivot if there is one.
Great post! Doesn't it make sense to load the index separately, instead of a single bundle? RN the client would bust it's cache every time the content changes?