These are chat archives for ethersphere/orange-lounge

15th
Sep 2017
Viktor Trón
@zelig
Sep 15 2017 07:16
@homotopycolimit would you also review this PR please ethereum/go-ethereum#15139 ? Is the behaviour in line what we discussed?
Nick Savers
@nicksavers
Sep 15 2017 07:53
@zelig you switched A and B in the swindle games document Figure 1 but not in the text. Does this mean that moving 'current payment balance' towards B the balance tilts in favor of A?
Viktor Trón
@zelig
Sep 15 2017 07:56
I guess it is ok like that no?
I would appreciate your thoughts on the paper @nicksavers :)
Daniel A. Nagy
@nagydani
Sep 15 2017 10:15
@VoR0220 no, it won't kill off anything. It's just a way to exclude some from deciphering swarm content.
Viktor Trón
@zelig
Sep 15 2017 12:19
@lmars got promise from @fjl to put review comments to the simulations PR today ... lets see
holisticode
@holisticode
Sep 15 2017 14:54
I usually run this before submitting a PR:
build/env.sh go run build/ci.go test
is this equivalent to what the Travis build does or is it not sufficient?
holisticode
@holisticode
Sep 15 2017 15:12
@nagydani "This would allow for using different encryption keys for each chunk"
does this indeed mean that if a file has size > chunksize, each chunk of such a file has a different enc key?
also, if we'd switch to the pyramid chunker, would that affect encryption performance?
holisticode
@holisticode
Sep 15 2017 15:18
(actually don't think so)
Daniel A. Nagy
@nagydani
Sep 15 2017 15:35
@holisticode yes, that is what it means. With stream ciphers, you cannot reuse the same key. However, the key would be part of the reference, so instead of the reference consisting of just the hash as in the case of unencrypted content, it will be hash + decryption key.
I don't think that chunker scheduling has any performance implications.
holisticode
@holisticode
Sep 15 2017 15:46
ok so the key is stored in the chunk
Daniel A. Nagy
@nagydani
Sep 15 2017 15:46
Correct. The key is stored in the chunk above it in the tree.
Right next to the hash.
holisticode
@holisticode
Sep 15 2017 15:47
this is the 4096 + 8 bytes?
Daniel A. Nagy
@nagydani
Sep 15 2017 15:47
4096+8 bytes is the chunk. This is the current chunk structure as well, there is no change here.
Except that we allow for shorter unencrypted chunks but encrypted chunks are uniformly 4104 bytes.
holisticode
@holisticode
Sep 15 2017 15:49
what I don't understand from the text is where the key is actually stored
Daniel A. Nagy
@nagydani
Sep 15 2017 15:50
Which key? For the root chunk, it is not stored anywhere, the client must know it to have access. For non-root chunks, it's next to the referencing hash.
holisticode
@holisticode
Sep 15 2017 15:50
yes non-root
so if a chunk is 4096, and its hash 8, where is the key'
?
Daniel A. Nagy
@nagydani
Sep 15 2017 15:51
Instead of containing 128 unencrypted hashes, non-root chunks are encrypted and contain 64 (hash,key) pairs
holisticode
@holisticode
Sep 15 2017 15:51
sorry for my ignorance
Daniel A. Nagy
@nagydani
Sep 15 2017 15:51
No, the 8 bytes are the length field. They encode the length of the underlying content.
holisticode
@holisticode
Sep 15 2017 15:51
ah yes
ah non-leaf chunks are encrypted hashes of their children
sorry I had lost this
thanks for your explanations
holisticode
@holisticode
Sep 15 2017 16:01
Instead of containing 128 unencrypted hashes, non-root chunks are encrypted and contain 64 (hash,key) pairs
maybe adding this to the text is a good idea?
on another note, I currently run manjao linux (arch based) - it's a rolling update distro
so it installed without me realizing go1.9
holisticode
@holisticode
Sep 15 2017 16:07
I am seeing issues with swarm and go1.9
i need further look though before reporting what it is (and if it's not related to other parts of the rolling update)
holisticode
@holisticode
Sep 15 2017 16:14
@zelig I will hold off from creating more diagrams until we have settled on a layout
holisticode
@holisticode
Sep 15 2017 16:32
@lmars what's your view on my report above regarding the snapshots?
Daniel A. Nagy
@nagydani
Sep 15 2017 17:26
Thanks for the feedback. I will add that text to the document's next revision (tonight)
holisticode
@holisticode
Sep 15 2017 17:32
I am seeing issues with swarm and go1.9
Probably related to my linux - disregard until further notice
Lewis Marshall
@lmars
Sep 15 2017 17:51
@holisticode sorry I've had a lot on the past few days, I'll take a look at your snapshot issue over the weekend
holisticode
@holisticode
Sep 15 2017 18:37
Thanks @lmars sorry didn't mean tu be pushy :)