protocol

Creating a new issue based on lbryio/lbry-sdk#877

• Update ensure_directory_exists() to check that the directory is writable by the current process.
• If it's not, return a useful error
• Add tests to cover the case when
  • download directory does not exist
  • download dir exists but is not writable
  • dir exists an

The htlcswitch test function initSwitchWithDB may create its own database if nil is passed in. Several tests that pass in nil never remove the temporary directory. I don't think any of the tests call db.Close so the test process will still have the file handle. This leads to too many open files... on my machine. The fix here is to remove the temp directory and call Close on the databas

We have DAEMON_START metric which shows duration and count of times we need to start up our own daemon. However, we do not have a metric to show how often there are users who are already running an ipfs daemon.

The webui is available when running the go implementation of IPFS, so if we see a lower count for this NO_DAEMON_START metric, then we could potentially redirect any users who are alread

Describe the bug

[3][50f8.50f8][10:08:10.721496][ api] Enter 7 (0x5622f42e4700).
[3][50f8.50f8][10:08:10.721690][ lib] ERROR, 336232638, "SSL_CTX_check_private_key failed".
[3][50f8.50f8][10:08:10.721699][ api] Exit 126
[3][50f8.50f8][10:08:10.721703][test] FAILURE - "/workspaces/msquic/src/test/lib/HandshakeTest.cpp":1945 - "ServerConfiguration.IsValid() not true"
[3][50f8.50f8][10:

I believe these are the knobs for data limits:

https://docs.rs/quinn/latest/quinn/struct.TransportConfig.html#method.stream_receive_window
https://docs.rs/quinn/latest/quinn/struct.TransportConfig.html#method.receive_window

Would like to be able to set these per-connection.

Also, would be great to release this: quinn-rs/quinn#1315

I suppose if a client violates

As seen in the Arweave integration PR the BlockPtr hash is limited to 32 bytes / 256 bits (H256 type).

This should be refactored so graph-node can deal with chains that have larger block hashes.

Seems that Firefox (around version 96-97) changed somethign and is now way more aggresive when a gateway does not start returning response immediately.

For example, first load (empty ipfs repo) of http://dist.ipfs.io.ipns.localhost:8080/go-ipfs/v0.11.0 in Firefox 97 produces:

![2022-02-11_19-09](https://user-images.githubusercontent.com/157609/153646056-9bc942ec-bec7-4f0a-b61b-53db7bf6896

I'm mocking a protocol with a function that looks like this:

func foo(
    completion: @escaping (
        _ bar: Int,
        _ baz: Int,
        _ qux: Int
    ) -> Void
)

The generated mocks do not properly remove the line breaks before attempting to put it into a single-line string literal, resulting in the errors Unterminated string literal, `Consecutive statement