@aboodman GraphQL problems such as joins, complex queries, versioning, too easy to overexpose data leading to very easy IDORs, File uploading, exposing business related errors, extremely hard to consume by another BE service, different AuthZ rules for different business functions, etc.
@aboodman Conceptually it seems very similar. I won't say GraphQL and this are opposite. In GraphQL, we define data to fetch from frontend using a query language. The schema is defined at backend. Zero looks like added BE impl to a query language.
@penberg Within a PR, we have multiple commits. When we merge PR, we squash and merge it. Main branch follows one commit per merged PR. PR still has history of all commits.
@_Kahzel@uwukko Either you don’t know c++/c or you don’t know TS/JS.
TS code transpiles to literal JS code. C++ codes does not gets transpiled to C code.
@annoyedcanuck1@uwukko TS gets transpiled into literal JS code, and then resulting JS code runs. Is this the same case with Scala? Does Scala code transpiles to Java code? No. It gets directly converted to bytecode.
@AwokeKnowing@uwukko So you are saying that c++ transpiles to actual C code first and then that C code gets compiled to assembly? Wow! Please don’t comment on things you don’t know.
@hetmehtaa the payload is passed inside "data" property. So, the thirdPart API should allow urlencoded content-type, and then decode it, and also accepts payload inside "data" property apart from normal response from OAuth APIs.
So, was the API allowing all these things?
@hetmehtaa Looks like I am missing something here. Can you please help in understanding it a bit better. In the code to submit data to thirdPart API, the payload would be submitted as "application/x-www-form-urlencoded" content-type. Also, even if API accepts urlencoded, ...
@ospfranco I just slap the compiler with clone every time it complains of borrowing. Rust runtime will figure out memory cleanup of borrowed and borrowing value. Keeps me sane. If I ever see a memory intensive path, then I will try correct borrowing.
@ludwigABAP Awesome! And then you have some guys who reads blogs on their videos and they have crazy subs and views. People are watching others read a blog on a video 😞
@matinzd There are still some Date related perf issues specific to Hermes. Date libraries don't work as fast in Hermes as they do in JSC or v8. So, anything you can do without libraries is better.
@matinzd Intl is now supported in Hermes. Although not whole API. But it has APIs we need. We were using formatter from date-fns and it was extremely slow. Also, don't construct new formatter for every string. Create one format at top level and just use .format on that formatter.
@BenLesh RxJS is nice and recently EffectTS is also talking about these things. But I like Sagas much better to handle these cases in imperative way. You can choose to ignore successive events, queue them, interrupt current run, have cancellation or rollback, etc.
@RaulJuncoV OR start with a Monolith and see what you want to break out. Or don't even need to breakout. You would still be deploying using docker and a container orchestration service. Microservices are hardly justifiable for teams size less than ~40.