Mixed-endian decoding?

[mhx]

I'm using fsst with great success to store large tables of filenames in a compressed read-only file system.

So far, the code base only supports little-endian hosts, but I'm trying to support big-endian in the future as well. I came across a few comments in the fsst code base, mostly in fsst_export, mentioning that "endian-conversion during hashing will be slow and self-defeating" and "Overall, we could support reconstructing an encoder for incremental compression, but should enforce equal-endianness. Bit of a bummer. Not going there now.".

For my use case, I only need to be able to construct a decoder from an "other-endian" symbol table.

I wonder what viable options exists here?

[mhx]

Never mind, I think I figured out a solution in #36.