automerge-repo-rs
automerge-repo-rs copied to clipboard
This commit has fixes for path partsing and it uses the root for temp files
I think it should at least be configurable where to store the temp files, but the same root as the data directory is safer, as you can't rename files across filesystem types in the way spanreed is doing so now.
I am also seeing filenames like this:
./259d/63393439333239642d366630662d346437612d613138312d636435323632386165626163/e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855.incremental
Which has 72 characters for the doc id and 64 for the hash. I haven't dug into why yet.
The loading function was also not previously working. The implementation is now functional, if not entirely correct.
@issackelly thanks for the contrib!
@alexjg I will let you review since you implemented this part.
This one doesn't address #52 in any meaningful way, we should extract that out of the issackelly:compaction branch. The character length issue is something like ::str::hex::str::hex::random_128_uint which is getting expected results. Fine to leave as is for now I guess but some optimizations for humans might be warranted