Complete the Jest DI with crypto #234
Conversation
src/setup/jest.ts
Outdated
| async getSymmKey(keyName: string, cfg?: Partial<Config>): Promise<CryptoKey> { | ||
| const mergedCfg = config.merge(this.cfg, cfg) | ||
| const maybeKey = this.store[keyName] | ||
| if(maybeKey !== null) { | ||
| return maybeKey | ||
| } | ||
| const key = await aes.makeKey(config.symmKeyOpts(mergedCfg)) | ||
| this.store[keyName] = key | ||
| return key | ||
| } |
There was a problem hiding this comment.
I duplicate some code here from keystore-idb.
We should probably eventually make it possible to use our dependencies in node. However, doing that for keystore-idb could be questionable, given it has "idb" in its name 😄
Any way, this feels kind of quick-and-dirty right now. Let me know if you have any other ideas.
Switch to using Uint8Arrays instead of node Buffers. This is something the js-ipfs folks apparently have gone through, too: ipfs/js-ipfs#3220
| @@ -36,7 +36,7 @@ | |||
| "prebuild": "rimraf dist && node scripts/gen-version.js", | |||
| "build": "tsc && rollup -c rollup.config.ts", | |||
| "start": "tsc -w", | |||
| "test": "jest", | |||
| "test": "jest --forceExit", | |||
There was a problem hiding this comment.
Hrm. This smells like there's something wrong with the tests... async operations not resolving?
There was a problem hiding this comment.
Yep. I tried to track it down for a while, but gave up. From my testing it seems like in some edge-cases it's possible that ipfs leaves a child process and some sockets hanging after await ipfs.stop(). Not sure what's up with that, but after fighting with it for a while I just gave up.
The IPFS we're using in the tests is basically in-memory only. It doesn't write to disk. So killing it shouldn't be an issue.
There was a problem hiding this comment.
Cool. yeah, something we can track down and fix as a follow up later, imo.
| @@ -1,8 +1,9 @@ | |||
| import crypto from "crypto" | |||
| import Ipfs, { IPFS } from "ipfs" | |||
| import { IPFS } from "ipfs-core" | |||
There was a problem hiding this comment.
Is this a size thing? (full IPFS includes more code than we use, etc?)
There was a problem hiding this comment.
Yep, exactly. It's just much less dependencies.
I think the difference is that ipfs also contains the CLI and an http-server, which ipfs-core does not.
Yeah. Weird. |
| async function runRoundTrip(message) { | ||
| const keyStr = await webnative.crypto.aes.genKeyStr() | ||
|
|
||
| const encoded = webnative.cbor.encode(message) |
There was a problem hiding this comment.
This test is failing for me with:
Evaluation failed: TypeError: Cannot read property 'encode' of undefined
There was a problem hiding this comment.
Did you yarn build before you yarn tested?
There was a problem hiding this comment.
I did not 🤦 and that was it. After building, this test and all the others are passing for me 💯
There was a problem hiding this comment.
Admittedly, it's really confusing. The puppeteer html file fixture uses the index.umd.min.js file, so unless one yarn builds before running the tests, it'll use an outdated webnative, but only for the puppeteer stuff.
This is preparation work for making it possible to write more sophisticated test, making test-driven-development for internal (non-browser-bundle-exposed) APIs possible.