Paul Dagnelie

My current theory on this is that there is a bug in `zfs_send_resume_impl` where the `raw` and `saved` flags aren't copied into the flags that are used for size estimation....

@awehrfritz It would be useful if we could get the output of the same commands I asked for in the other thread (https://github.com/zrepl/zrepl/issues/463#issuecomment-866192806); obviously replace the dataset names with the...

Sorry, that's correct. The *feature* is called `is_some_with`, which is what confused me. EDIT: Updated the original comment.

This similarly makes it impossible to use mold to build the ZFS kernel modules. Perhaps it is possible to replicate the functionality of the linker script using objcopy or other...

1. Okay, but in that case, why not have a mechanism to have the default behavior for every other data type? As it is, there is no simple way to...

Would that allow us to create a contiguous chunk of space that btrfs wouldn't try to read to or write from? I assumed that they behaved similarly to ZFS's snapshots,...

We allocate the buffer for the embedded block on the stack, and we don't zero it before use. The unpacking code also doesn't zero anything after the tail. We send...

We just ran into this internally, and @rincebrain's analysis is correct. The proper fix here probably involves tweaking ax_python_devel.m4 to use a scheme that puts the file in the right...

https://gist.github.com/pcd1193182/f177c2899219ad874fa45e61a6cf3d3b Might be a starting point for a fix?

I do think it's probably worth considering intermediate sizes between 4k and 128k; losing compression benefits is certainly bad, but a 1:8 record:sector size ratio from before was clearly treating...