News for hackers. Stuff that matters.
Re:An introduction to IPFS by XorNot on Saturday February 20, 2016 @03:50AM (#11139434)
The problem with this and other initiatives is they don't really have a good story for what happens with a lot of data-lint regular people have. They make broad assumptions that we have near limitless storage resources (including those needed for redundancy) when private users definitely don't have that and even at enterprise levels, the story is still fairly complicated.

Immutable is an interesting idea - it's a lot less interesting when 100 different copies of the same slightly changed RAW file from my digital camera are using up 100s of gigabytes. Or I misclick and something goes into the public store which shouldn't - I might not be able to get rid of all of it, but I should be able to undo it a little.

Re:An introduction to IPFS by rakoo on Saturday February 20, 2016 @04:18AM (#11139493)
If you look a little bit closer into camlistore, you'll see that it splits content with content-defined chunking, ie when there's a change in the middle of the file only this zone (camlistore targets 8kB) is really added to camlistore; there's an indirection between the file as an object and the actual content. That means that hundreds (or even thousands) of slight changes to the same big file won't change much.

Regarding misclicks, for camlistore everything is private by default; in order to make something public you have to build an authorization, which you give to someone. You also have the possibility to remove that authorization, and the data won't be publicly accessible anymore.

Re:An introduction to IPFS by XorNot on Saturday February 20, 2016 @05:02AM (#11139578)
That doesn't address the issue: any compressed content (and raw files are losslessly compressed) tends to really break rolling hash type splitting systems.