The way the BCacheFS situation has been playing out is a tragedy. I had very high hopes for it.

InsideOutSanta 2 days ago | parent | next [-]

Yeah, this all seems so unnecessary. I hope Kent can either figure out how to work in the context of a larger team or find somebody who can do it on his behalf.

▲

Voultapher 2 days ago | parent | next [-]

Don't hold your breath.

I've approached Kent two weeks ago and offered to help upstream changes into the kernel so he wouldn't have to interact with any of the kernel people ... he claimed without being able to explain why that having a go-between couldn't possibly help and that if he couldn't dictate the kernel release schedule he wouldn't want to ship the software anyway. And then proceeded to dunk on btrfs.

	▲	yencabulator 16 hours ago \| parent [-]
		Totally expecting to see bcacheos next. (Then maybe down the line people will realize Linus had a heart of gold behind his foul mouth.)

▲

johnisgood 2 days ago | parent | prev [-]

> Once the BCacheFS maintainer behaves [...]

So, there are still behavioral issues here I take it? That is a bummer. This is not news to me, but I thought the situation has changed ever since.

▲

motorest 2 days ago | parent [-]

> So, there are still behavioral issues here I take it?

From the past discussion, it's mainly grave behavioral issues but they also end up being technical. Such as trying to push new untested features into RCs and breaking builds, and resorting to flame wars to push these problematic patches forward instead of actually working them out with maintainers.

But yeah, the final straw was a very abusive email sent to a maintainer in the mailing list.

	▲	koverstreet a day ago \| parent [-]
		...Where are you getting this stuff? Seriously, every time I read tales about all the horrific things I've done in the kernel community, the stories grow and get wilder and wilder. Oh wait. It's you.

▲

johnisgood 2 days ago | parent | prev [-]

Same. I liked many of its features (actually, all features, see https://bcachefs.org) and I was waiting for it to become usable, but I guess that day will never come now?

So, the alternative is ZFS only, maybe HAMMER2. HAMMER2 does not look too bad either, except you need DragonflyBSD for that.

▲

ahartmetz 2 days ago | parent | next [-]

What I expect to happen is that bcachefs stabilizes outside of mainline, and after that, it can be merged back because no large patches = not much drama potential.

▲

yjftsjthsd-h 2 days ago | parent [-]

My concern is that historically some of the conflict over bcachefs was because work on it touched code outside the filesystem area. If a year from now they show up and say "okay, here's the new bcachefs code that's 100% formally verified and has zero bugs and is good to go!", then it still could fail to get merged because they rewrote parts of (say) the block device system and the maintainers of that part of the kernel don't like the changes.

	▲	yencabulator 16 hours ago \| parent \| next [-]
		Hey, the block i/o subsystem is at least connected! Kent also wrote a brand-new locking system: https://lwn.net/Articles/755164/
	▲	ahartmetz a day ago \| parent \| prev [-]
		I didn't forget about these things, but I was under the impression that they had been resolved one way or another.

▲

ThatPlayer 2 days ago | parent | prev [-]

It's not unusable, I use it on a spare computer for fun, cuz I want tiering of SSD + HDDs. And this doesn't mean development has stopped, just not done in the kernel.

▲

johnisgood 2 days ago | parent [-]

True, I did not mean to say usable, I meant to say that "officially supported", but I will give it a go with a custom built kernel sooner or later.

For how long have you been using it? Any issues? Any favorite feature(s) in particular?

	▲	ThatPlayer a day ago \| parent [-]
		I've been using it about a year or so. Like I said, the 'killer feature' for me is tiering. I'm not using it on my main computer, but a spare parts computer with whatever storage I have laying around thrown into it, so being able to combine a 3TB HDD with a 1/2 TB SSD. All the storage stays usable, and I get better performance. Not perfect though. Performance at times doesn't seem better than just SSD, though I didn't really test that and I'm not sure if it's just hitting the HDD at those times. And there was a kernel version that didn't boot, though I'm staying on the bleeding edge with Arch Linux.