Why is crypto.subtle.digest
designed to return a promise?
Every other system I’ve ever worked with has the signature hash(bytes) => bytes
, yet whatever committee designed the Subtle Crypto API decided that the browser version should return a promise. Why? I’ve looked around but I’ve never found any discussion on the motivation behind that.
looks
https://developer.mozilla.org/en-US/docs/Web/API/SubtleCrypto/digest
Well, it can’t take a stream as input, so it’s not that an output can’t be provided because the input might not be fully known.
And it seems kind of odd if the aim is parallel execution.
pokes around a bit more
I don’t really code in Javascript much, but as I very vaguely recall, at least in browsers, Javascript doesn’t normally have the ability to do concurrent execution…there’s some sort of mechanism that isolates code running in parallel, Web Workers, as I recall.
One of the things that Mozilla’s docs on Promise mentions is this:
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Promise
So, one thing that you normally have to do in APIs that have cooperative multithreading going on is, if you want the system to stay responsive, is to make sure that if you’re gonna yield to other threads if you’re gonna be doing some work for an extended period of time. Like, say you’ve got something going on that’s CPU-bound and something that’s network-bound…you don’t want to have one blocking on the other. You want to slice up the tasks and switch back and forth between them to keep the network connection saturated.
I am thinking that it’s possible that the goal here is to do that. Like, you might want to be generating a hash of a big block of data, say a couple gigs. Maybe it takes Javascript a while to do that. So you generate a Promise for that computation, along for the other things you need to do, and then wait for any of them to complete.
If you don’t have anything else going on in your particular codebase, then you can just immediately block until the Promise is fulfilled.
That being said, that’s just my kneejerk reaction based on about a two-minute skim and some familiarity with past systems that have worked in kinda similar ways.
To add - blocking the main thread on a long running task in the browser can make the page unresponsive. There’s not really a way, as far as I know, to “block until a promise completes”, which might be the source of the frustration. It seems to me that was intentional by the ones who designed this function.
That seems like a good guess, I can see why async hashing could be useful. But it would be nice if there was an alternative API that was blocking so my code wouldn’t get infected with async/await all over the place…