qri-io / cafs Goto Github PK

cafs is now a subpackage of qri-io/qfs. We should first make sure qfs includes all the same CI checks as this repo, update this repo's readme to point viewers to cafs, and archive this repo.

casters need to be able to add a "directory" of byte slices. Maybe:

type Node struct {
   Name string
   Data []byte
   Children []*Node
}

type castore interface {
   ...
   AddTree(node *Node) error
}

from there we could add helper methods that convert things like .zip archives & unix directories into a tree of nodes.

In the dataset package, we added VisConfig as a field to Dataset. When we try to write to the filestore, however, the adder only allows for 8 channels. We need to increase this to 9 in order to allow all the Dataset fields to be written to the filestore.

This needs to be changed in the cafs/ipfs and cafs/memfs packages

This causes annoying behavior: no error, but hanging! Maybe we need a function that returns whether all the channels are full? Not enough knowledge to make this function, also it's 1:30am, lol

Either our implementation of the Adder interface or the way we're consuming an adder isn't behaving properly. the size of the buffer theoretically shouldn't matter But clearly the buffer needs to be large enough to accommodate all files before reading starts, which is not how this code is supposed to behave. Or at least it should be possible to write code that doesn't behave this way.

Let's write a test that adds more files to adder than the size of the adder channel buffer and make it not hang forever.

I'm leaving this as a stub for later and will add steps to complete in the future. If you'd like to work on this issue and would like more guidance, comment here & I'd be happy to add details!

castore.Put(data []byte) error -> castore.Put(data []byte, pin bool) error