feat: Add support for specifying hash algorithms in files.add #1308
Conversation
ghost
assigned 
| 'keccak-256', | ||
| 'keccak-384', | ||
| 'keccak-512' | ||
| ] |
| const runOnAndOff = require('../utils/on-and-off') | ||
|
|
||
| // TODO: Test against all algorithms Object.keys(mh.names) |
daviddias
deleted the
feat_add-support-for-specifying-hash-algorithm-in-files-add
branch
|
Hey @vasco-santos. I've been working with this this weekend and have run into a few pieces that I'd love your thoughts on: go/js ipfs hash differences: DAG traversal We can request the specific DAGNode if we know it's hash but it's not possible to traverse the DAGLinks of something added with a non-default algorithm. Seems like IPLD isn't aware that the links are of a different alg, they may be casting all hashes to Do you know what's wrong here? Wanted to run this by you before creating an issue. |
|
Hey @JonKrone Thanks for reaching out. I tested across encoding + decoding and it seemed good. However, I tested your example, as well as using other hash algorithms, but the problem persists. At a first sight, your suggestion seems to be the problem. Would you mind to create the issue and verify it? |
Since PR 1005 has been blocked for a while, I went through it.
As the PR origin is from an external organization, I created a new branch instead of rebasing it in its origin. Shortly, I got the @alanshaw (thanks :D ) implementation with
js-ipfsmaster branch and made a few changes, as a consequence of the modifications that occurred meanwhile incomponents/objects, as well as inresourcesand a minor refactor.Finally, I added tests for several hash algorithms, since there are more than 300. Should I add more to the tests?