We're moving forward as always. Latest features would include distributed pubsub, filestore (allows you to add files without duplicating them) and interop between the browser and desktop nodes. Any specific part you're looking at?
1) What's the status of (supported as a real feature, not just manually changing the bootstrap nodes and hoping everyone else does too) private IPFS networks? If it's there already, how stable is its configuration (i.e. if I get my friends on a private IPFS network will I likely have to get them all to update a bunch of config in 6 months or a year)?
2) Does filestore also let you store, say, newly pinned files in your regular file tree? That is, can you pin a hash for a file (or tree) you don't already have and provide an ordinary file system location where it should go when it's downloaded? Or do you have to copy it out of IPFS' normal repo manually, then re-add it in the new location? Also: how does filestore behave if files are moved/deleted?
3) What rate of repo changes requiring upgrades can we expect for the future? That is, how stable is the current repo structure expected to be? Is the upgrade process expected to improve and/or become automated any time soon?
4) Is there a table of typical resource requirements somewhere? I'm looking for "if you want to host 10TB and a few 10s of thousands of files, you need a machine X GB of memory. If you want to host 500MB, you only need Y GB of memory. If you have 2TB but it's in many, many small files, you need Z GB of memory", or else a formula for a achieving a best-guess for that. For that matter, how predictable is that at this point?
The use case I've been excited to use IPFS for since I found out about it is a private, distributed filesystem for my friends and family. Easy automated distributed backups/integrity checking on multiple operating systems, access your files at someone else's house easily, that sort of thing. Filestore finally landed, which was a big piece of the puzzle (the files have to remain accessible to ordinary tools and programs or I'll never get buy-in from anyone else), so that's exciting. Now I'm just waiting for docs to improve (so I'm not searching through issue notes to learn which features exist and how to use them) and for a sense that it's stable enough that I won't be fixing brokenness on everyone's nodes several times a year.
