DEV Community

Cover image for Javascript Promises in depth with V8 engine internals
server side digest
server side digest

Posted on

Javascript Promises in depth with V8 engine internals

😅 Being a single threaded language most of the people think that Javascript or NodeJS is slow and not scalable. Python and JS be like:-

Javascript Python meme

Well, it depends. If your service is I/O (input output like file reading, DB call, Http call etc) heavy which means it has dependency on other resource being read or written.

⚡️ There, NodeJS performs good as compared to other stacks in general when your service is IO heavy instead of CPU heavy.

❓ This is because NodeJS or JS has something like handing over its work to the underlying thread of the Operating system to carry out IO execution.

It passes the callback (a function) to that thread and continue with the execution of other instructions in the code, which is called async programming also in fancy terms.

Well Well Well, Let's take a step back and understand what we wrote above:-

Blocking nature of JS

JS blocking nature

  • Javascript is a single threaded language and hence can work on one task at a time unlike Java where multiple threads are present to handle workload in parallel (multithreading).
  • So, if one function/instruction is running, next instruction will wait for this instruction to get completed and we can't run them in parallel.

Let's see how one instruction blocks another one:-

Step 1 of blocking nature of js

Step 2 of blocking nature of js

Step 3 of blocking nature of js

Step 4 of blocking nature of js

✅ You can refer video explanation (in english) with code examples and v8-engine internals here:- https://www.youtube.com/watch?v=JN89L2SqPA8

Non blocking nature

Now, in JS runtime environment (NodeJS) we have some C++ APIs that have some functions like setTimeout, Promises, localStorage which are executed by the threads of our Operating systems, not by the main JS thread.

You can see the right box having C++ APIs here:-

C++ Nodejs APIs

  • Whenever, we perform IO through any C++ API like executing a promise , performing HTTP call or DB operation etc JS main thread hands over the main work to the OS thread and the main thread of JS continues working on other code and doesn't wait for the Async code to run.
  • It provides a callback (a function) to execute when call stack gets empty (call stack has only the sync code and async code goes to microtask or task queue)
  • Event loop, loops through these queues when call stack gets empty & put the callbacks from these queues to the main stack and returns the result.

Promise execution in task queue

setTimeout execution

For more in depth details of How promises are implemented in C++ code, refer this:- https://www.youtube.com/watch?v=JN89L2SqPA8


🔥 Follow for more such articles...

Top comments (7)

Collapse
 
dev_sd profile image
Soumajit Das

Nice explanation

Collapse
 
light_seekern profile image
Taha Syed

Nice presentation

Collapse
 
darkwiiplayer profile image
𒎏Wii 🏳️‍⚧️

Using async I/O isn't really a node-specific thing though. Never has been. Even less true today though, with many other languages implementing their own flavours of async I/O with many different mechanisms.

Personally, I think JS ultimately chose one of the less convenient abstractions for this sort of behaviour. Languages that opted for coroutines like Lua or more recently Ruby are a lot easier to use and don't suffer from the coloured functions problem.

Collapse
 
sunilkhadka001 profile image
Sunil khadka

I expected to see promise internals like the synchronous nature of executor function and then resolve and reject but the article was mostly about event loop. Anyway, great effort and I learned a few new stuff.

Collapse
 
monica_exists profile image
Monica Blake

Great interview prep stuff. Some images are too big though.

Why do the other two comments look like bot comments? 🤣

Collapse
 
edwin_alaekpere_d531fa083 profile image
Edwin Alaekpere

Cool but not really as in depth as you said

Collapse
 
kishan_rajput_d882523cb08 profile image
Kishan Rajput

Which memory async code executed