Akira Hayakawa

An other idea is ack the deferred flush requests with error (EIO or DM_ENDIO_REQUEUE). But still the inflight ios may needs a rambuf to be flushed to acquire new rambuf...

this is unlikely to occur but if we don't discard the segment everytime, the SSD controller can never realize some blocks in the segment are reclaimable because there isn't overwrite...

but please be sure that no bad impact on the performance. This should be easily measurable by randwrite test

If we do this, we should insert blkdev_issue_discard into ``` c void acquire_new_seg(struct wb_device *wb, u64 id) { struct segment_header *new_seg = get_segment_header_by_id(wb, id); /* * We wait for all...

Vagrant because the testbed can be made inside travis CI so we can use the CI infra to test dm-writeboost on every PR

The Vagrant VM is really helpful for bug reporters to reproduce their issue. The VM should be dmts installed so a reporter can write a dmts test case and send...

bytes crate also supports serde features. IIUC, those two has similar functionality. Which is the recommendation of serde community?

I want this feature because `multi_cartesian_product` is too slow.

What if we want `Serialize` and `Deserialize` for `HashMap`? I think the http-serde doesn't work in this case but we need a wrapper around `Uri`. This way, the application code...

@GothAck Here is the documentation https://akiradeveloper.github.io/lol/gateway.html but it may be a little unkind for new comers. Gateway is a kind of Tonic's [Channel](https://docs.rs/tonic/latest/tonic/transport/struct.Channel.html) but automatically directs the client request to...