Is it possible to 'ipfs get' without save things twice?

Help
1

I try to replicate my data in a number of computers behind different internet connections, so, access from other people will be faster.

To replicate my data I just run:
ipfs get /ipns/k51aaaa…

But this reserves storage in folder
~/.ipfs/blocks
as individual scattered blocks
and in folder ~/k51aaaa…
as file hierarchy.

Because in secondary computers I don’t care about availability of data in meaningful state, is there a way to avoid the creation of folder ~/k51aaaa…?
Even obscure method will be fine (even -if possible- to download somehow block-by-block)

Or, else, is possible to avoid ~/.ipfs/blocks and with some hack on filestore use directly ~/k51aaaa…?

2

Soon…

Currently no but I’m working on reflinking flat FS (using the copy range reflink of the latest linux FS (like btrfs) to store multiple files with the same data blocks).

1 Like
3

BTW if you are willing to implement that yourself you could hook in the block exchange layer.

4

You can just download just the blocks without creating an output folder with ipfs refs -r /ipns/k51aaaa....

1 Like
5

if streaming them is a valid option I guess ipfs cat does the same in a simpler way.

6

ipfs cat would do the same for a single file. refs -r allows to download a full folder structure.

Iirc ipfs pin add could be faster due to paralel dag traversing (but in this case it would need to be followed by pin rm).

2 Likes