Eric Biggers

No, storing those fields is not implemented yet. I think they are rarely needed, actually. Why do you need them?

First, note that the `silesia` corpus is not great, especially when it is compressed all as one file, as that gives very different weights to the files inside it. It's...

> I still stand by my opinion regarding levels 6 - 9. Even with your odd data sample it does a good job, better than current parameters; only level/s 10+...

There have been a lot more requests for CMake, including an existing partially-complete pull request (https://github.com/ebiggers/libdeflate/pull/101), so that's the one I'd probably go with if I were to change to...

I haven't used this compiler before, and it doesn't appear to be commonly used. It also doesn't appear to be easily installable on Linux like `apt-get install clang` (presumably because...

Which compression level are you using? I think the cutoff to skip compression shouldn't be fixed at 16 bytes, but rather it should depend on the compression level. The lower...

Well, it's easy to see the performance difference in decompression if you have a BMI2-capable processor. Do a benchmark with the current code, then change `xpack_decompress.c` to always call `xpack_decompress_default()`...

Large buffer sizes aren't supported very well yet; I wouldn't really suggest anything over 2^19 (524288) bytes. This is definitely a big problem and I am planning to address it.

@gmarcais, can you confirm whether the silent update of the release tarball is intentional?

@dvyukov is there any update on this? There's also interest in enabling the `-E encoding=utf8` mkfs option so that the ext4 Unicode / casefolding support in kernel v5.2+ will be...