GitLab

fix(test): fix a flaky pubsub test

README: stub

test: test ReadAt if implemented

(I plan on adding support to the http client, at least)

test: fix put with hash test

We just changed ID/"id" to IDENTITY/"identity" to match the multicodec table and
avoid confusion with peer IDs, CIDs, etc.

Unfortunately, this breaks the `--hash=id` flag. Luckily, I'm pretty sure
nobody's actually using this. As putting a block with an identity hash is
useless. If they are users, we can go about adding in backwards compatibility
hacks later.

dep: update libp2p again

Bump go-libp2p-core, up test key size to 2048

migrate to go-libp2p-core.

tests: expose TestSuite

switch to base32 cidv1 by default

path: drop error from ParsePath

tests: fix a bunch of small test lints/issues

And fix for peer ID formatting changes.

fixes https://github.com/ipfs/go-ipfs/pull/6222#issuecomment-483479039

Update Pin.RmRecursive docs to clarify shared indirect pins are not removed

don't close the top-level addr

fix a bunch of small test "bugs"

See https://github.com/ipfs/go-ipfs-http-client/pull/10/files#r269268326

Otherwise, we can get random panics form dereferencing nil pointers.

remove Wrap

This can be trivially implemented by the end-user if desired. The best the
CoreAPI can do is name the file with it's own hash so this isn't really all that
helpful either.

Note: This differs from js-ipfs because _there_, all files have paths (even
outside directories).