Automatic Updates

[davidar]

All of the archives are currently being imported to IPFS manually. This is fine as a starting point, but we need to write some scripts to keep them up to date with the origin, run them periodically and publish the changes over IPNS.

[davidar]

More specifically, what I'm looking for is:

• [ ] a standard format for storing archive metadata: authoritative timestamps, raw file hashes (e.g. md5sums/sha1sums, not just unixfs hashes), and whatever other information is useful in determining if a file has been updated
• [ ] protocal handlers (rsync, oai, http, ftp, resync, etc) that can use this metadata to determine exactly which files need to be updated
• [ ] a tool that can splice the new files into the existing archive, without having to download the current contents of the archive (which could be quite large)
• [ ] a script to glue all of this together nicely, that can automatically push updates to IPNS, and is cron-friendly

CC: @rht @jbenet #18

[jbenet]

very much agreed. ideally for every thing we import we would create a program to do it, that updates with a new root when run again, (and the ipfs root of the archive would make sure to link to said program too)

i think we can write a standard format for commandline tools to do this:

> archive-arxiv.org <last-root>
> archive wikipedia.org <last-root>

etc.

[davidar]

@vinctux @CounterPillow @jbenet Migrated off-topic discussion to ipfs/notes#50

[davidar]

In which language the update tools should be written?

@vinctux my personal preference for this would be Python :)