EROFS File-System Adding LZMA Compression Support Via New MicroLZMA

[xhustler]

Better performance, less read amplification, faster lookup, no additional allocations (aside from some decompression contexts), more compact layout. Yes, its easily noticeable if you boot a live system from USB2 or use some cache-constrained cpus (Arm, Atom).
Its not really surprising, given that it was developed after squash wasn't good enough, benefit of hindsight and that stuff.

Apart from embedded, it seems the next big target are container images.

Thanks for this. Any idea how EROFS performs when coupled with AUFS?

[discordian]

For now, I will stick with squashfs given it's stability and maturity. What advantages does erofs have over squashfs? The erofs page talks about performance vs speed - has any one noticed this IRL?

Better performance, less read amplification, faster lookup, no additional allocations (aside from some decompression contexts), more compact layout. Yes, its easily noticeable if you boot a live system from USB2 or use some cache-constrained cpus (Arm, Atom).
Its not really surprising, given that it was developed after squash wasn't good enough, benefit of hindsight and that stuff.

Apart from embedded, it seems the next big target are container images.

[xhustler]

No reason to not use erofs in place of squash everywhere (a recent kernel is available).
I think zstd would've been the better choice, but I guess lzma gives some better bulletpoints (smallest size)

For now, I will stick with squashfs given it's stability and maturity. What advantages does erofs have over squashfs? The erofs page talks about performance vs speed - has any one noticed this IRL?

[arQon]

Is is snaps that use squashfs, or flatpak, or both?

Regardless, discordian is right: by the time your de/compressor outperforms an alternative on literally every metric, it's time to throw the old one away.

[piotrj3]

This is interesting file system, naked out of all needless headers, truly eroticFS

[discordian]

I think I may have missed the news on this, what's the problem it's solving over squashfs?

With the now added larger blocks and a high compression rate: Being better at everything squashfs does.

[jntesteves]

No reason to not use erofs in place of squash everywhere (a recent kernel is available).

I think I may have missed the news on this, what's the problem it's solving over squashfs?

[hsiangkao]

Yo dawg, I heard you liked file compressors so I made a file compressor for your file compressor so you can compress what you've compressed.

Nope, which means some part of the file is compressed with one algorithms, but the other part of the file is compressed with other else.

[discordian]

Is EROFS used anywhere outside of Android?

No reason to not use erofs in place of squash everywhere (a recent kernel is available).
I think zstd would've been the better choice, but I guess lzma gives some better bulletpoints (smallest size)

[skeevy420]

...even used for secondary compression on top of any other primary compression algorithm for this read-only file-system.

Yo dawg, I heard you liked file compressors so I made a file compressor for your file compressor so you can compress what you've compressed.

Is EROFS used anywhere outside of Android?